/src/gdal/ogr/ogrsf_frmts/geojson/ogrgeojsonutils.cpp
Line | Count | Source (jump to first uncovered line) |
1 | | /****************************************************************************** |
2 | | * |
3 | | * Project: OpenGIS Simple Features Reference Implementation |
4 | | * Purpose: Implementation of private utilities used within OGR GeoJSON Driver. |
5 | | * Author: Mateusz Loskot, mateusz@loskot.net |
6 | | * |
7 | | ****************************************************************************** |
8 | | * Copyright (c) 2007, Mateusz Loskot |
9 | | * Copyright (c) 2010-2013, Even Rouault <even dot rouault at spatialys.com> |
10 | | * |
11 | | * SPDX-License-Identifier: MIT |
12 | | ****************************************************************************/ |
13 | | |
14 | | #include "ogrgeojsonutils.h" |
15 | | #include <assert.h> |
16 | | #include "cpl_port.h" |
17 | | #include "cpl_conv.h" |
18 | | #include "cpl_json_streaming_parser.h" |
19 | | #include "ogr_geometry.h" |
20 | | #include <json.h> // JSON-C |
21 | | |
22 | | #include <algorithm> |
23 | | #include <memory> |
24 | | |
25 | | const char szESRIJSonFeaturesGeometryRings[] = |
26 | | "{\"features\":[{\"geometry\":{\"rings\":["; |
27 | | |
28 | | // Cf https://github.com/OSGeo/gdal/issues/9996#issuecomment-2129845692 |
29 | | const char szESRIJSonFeaturesAttributes[] = "{\"features\":[{\"attributes\":{"; |
30 | | |
31 | | /************************************************************************/ |
32 | | /* SkipUTF8BOM() */ |
33 | | /************************************************************************/ |
34 | | |
35 | | static void SkipUTF8BOM(const char *&pszText) |
36 | 903k | { |
37 | | /* Skip UTF-8 BOM (#5630) */ |
38 | 903k | const GByte *pabyData = reinterpret_cast<const GByte *>(pszText); |
39 | 903k | if (pabyData[0] == 0xEF && pabyData[1] == 0xBB && pabyData[2] == 0xBF) |
40 | 239 | pszText += 3; |
41 | 903k | } |
42 | | |
43 | | /************************************************************************/ |
44 | | /* IsJSONObject() */ |
45 | | /************************************************************************/ |
46 | | |
47 | | static bool IsJSONObject(const char *pszText) |
48 | 881k | { |
49 | 881k | if (nullptr == pszText) |
50 | 0 | return false; |
51 | | |
52 | 881k | SkipUTF8BOM(pszText); |
53 | | |
54 | | /* -------------------------------------------------------------------- */ |
55 | | /* This is a primitive test, but we need to perform it fast. */ |
56 | | /* -------------------------------------------------------------------- */ |
57 | 1.39M | while (*pszText != '\0' && isspace(static_cast<unsigned char>(*pszText))) |
58 | 513k | pszText++; |
59 | | |
60 | 881k | const char *const apszPrefix[] = {"loadGeoJSON(", "jsonp("}; |
61 | 2.64M | for (size_t iP = 0; iP < sizeof(apszPrefix) / sizeof(apszPrefix[0]); iP++) |
62 | 1.76M | { |
63 | 1.76M | if (strncmp(pszText, apszPrefix[iP], strlen(apszPrefix[iP])) == 0) |
64 | 1.57k | { |
65 | 1.57k | pszText += strlen(apszPrefix[iP]); |
66 | 1.57k | break; |
67 | 1.57k | } |
68 | 1.76M | } |
69 | | |
70 | 881k | if (*pszText != '{') |
71 | 789k | return false; |
72 | | |
73 | 92.2k | return true; |
74 | 881k | } |
75 | | |
76 | | /************************************************************************/ |
77 | | /* GetTopLevelType() */ |
78 | | /************************************************************************/ |
79 | | |
80 | | static std::string GetTopLevelType(const char *pszText) |
81 | 41.3k | { |
82 | 41.3k | if (!strstr(pszText, "\"type\"")) |
83 | 19.5k | return std::string(); |
84 | | |
85 | 21.7k | SkipUTF8BOM(pszText); |
86 | | |
87 | 21.7k | struct MyParser : public CPLJSonStreamingParser |
88 | 21.7k | { |
89 | 21.7k | std::string m_osLevel{}; |
90 | 21.7k | bool m_bInTopLevelType = false; |
91 | 21.7k | std::string m_osTopLevelTypeValue{}; |
92 | | |
93 | 21.7k | void StartObjectMember(const char *pszKey, size_t nLength) override |
94 | 100k | { |
95 | 100k | m_bInTopLevelType = false; |
96 | 100k | if (nLength == strlen("type") && strcmp(pszKey, "type") == 0 && |
97 | 100k | m_osLevel == "{") |
98 | 14.4k | { |
99 | 14.4k | m_bInTopLevelType = true; |
100 | 14.4k | } |
101 | 100k | } |
102 | | |
103 | 21.7k | void String(const char *pszValue, size_t nLength) override |
104 | 42.2k | { |
105 | 42.2k | if (m_bInTopLevelType) |
106 | 13.5k | { |
107 | 13.5k | m_osTopLevelTypeValue.assign(pszValue, nLength); |
108 | 13.5k | StopParsing(); |
109 | 13.5k | } |
110 | 42.2k | } |
111 | | |
112 | 21.7k | void StartObject() override |
113 | 42.8k | { |
114 | 42.8k | m_osLevel += '{'; |
115 | 42.8k | m_bInTopLevelType = false; |
116 | 42.8k | } |
117 | | |
118 | 21.7k | void EndObject() override |
119 | 21.7k | { |
120 | 11.4k | if (!m_osLevel.empty()) |
121 | 11.4k | m_osLevel.pop_back(); |
122 | 11.4k | m_bInTopLevelType = false; |
123 | 11.4k | } |
124 | | |
125 | 21.7k | void StartArray() override |
126 | 37.7k | { |
127 | 37.7k | m_osLevel += '['; |
128 | 37.7k | m_bInTopLevelType = false; |
129 | 37.7k | } |
130 | | |
131 | 21.7k | void EndArray() override |
132 | 29.4k | { |
133 | 29.4k | if (!m_osLevel.empty()) |
134 | 29.4k | m_osLevel.pop_back(); |
135 | 29.4k | m_bInTopLevelType = false; |
136 | 29.4k | } |
137 | 21.7k | }; |
138 | | |
139 | 21.7k | MyParser oParser; |
140 | 21.7k | oParser.Parse(pszText, strlen(pszText), true); |
141 | 21.7k | return oParser.m_osTopLevelTypeValue; |
142 | 41.3k | } |
143 | | |
144 | | /************************************************************************/ |
145 | | /* GetCompactJSon() */ |
146 | | /************************************************************************/ |
147 | | |
148 | | static CPLString GetCompactJSon(const char *pszText, size_t nMaxSize) |
149 | 79.3k | { |
150 | | /* Skip UTF-8 BOM (#5630) */ |
151 | 79.3k | const GByte *pabyData = reinterpret_cast<const GByte *>(pszText); |
152 | 79.3k | if (pabyData[0] == 0xEF && pabyData[1] == 0xBB && pabyData[2] == 0xBF) |
153 | 100 | pszText += 3; |
154 | | |
155 | 79.3k | CPLString osWithoutSpace; |
156 | 79.3k | bool bInString = false; |
157 | 447M | for (int i = 0; pszText[i] != '\0' && osWithoutSpace.size() < nMaxSize; i++) |
158 | 447M | { |
159 | 447M | if (bInString) |
160 | 197M | { |
161 | 197M | if (pszText[i] == '\\') |
162 | 322k | { |
163 | 322k | osWithoutSpace += pszText[i]; |
164 | 322k | if (pszText[i + 1] == '\0') |
165 | 221 | break; |
166 | 322k | osWithoutSpace += pszText[i + 1]; |
167 | 322k | i++; |
168 | 322k | } |
169 | 196M | else if (pszText[i] == '"') |
170 | 3.41M | { |
171 | 3.41M | bInString = false; |
172 | 3.41M | osWithoutSpace += '"'; |
173 | 3.41M | } |
174 | 193M | else |
175 | 193M | { |
176 | 193M | osWithoutSpace += pszText[i]; |
177 | 193M | } |
178 | 197M | } |
179 | 250M | else if (pszText[i] == '"') |
180 | 3.45M | { |
181 | 3.45M | bInString = true; |
182 | 3.45M | osWithoutSpace += '"'; |
183 | 3.45M | } |
184 | 246M | else if (!isspace(static_cast<unsigned char>(pszText[i]))) |
185 | 209M | { |
186 | 209M | osWithoutSpace += pszText[i]; |
187 | 209M | } |
188 | 447M | } |
189 | 79.3k | return osWithoutSpace; |
190 | 79.3k | } |
191 | | |
192 | | /************************************************************************/ |
193 | | /* IsGeoJSONLikeObject() */ |
194 | | /************************************************************************/ |
195 | | |
196 | | static bool IsGeoJSONLikeObject(const char *pszText, bool &bMightBeSequence, |
197 | | bool &bReadMoreBytes, GDALOpenInfo *poOpenInfo, |
198 | | const char *pszExpectedDriverName) |
199 | 450k | { |
200 | 450k | bMightBeSequence = false; |
201 | 450k | bReadMoreBytes = false; |
202 | | |
203 | 450k | if (!IsJSONObject(pszText)) |
204 | 416k | return false; |
205 | | |
206 | 33.6k | const std::string osTopLevelType = GetTopLevelType(pszText); |
207 | 33.6k | if (osTopLevelType == "Topology") |
208 | 68 | return false; |
209 | | |
210 | 33.5k | if (poOpenInfo->IsSingleAllowedDriver(pszExpectedDriverName) && |
211 | 33.5k | GDALGetDriverByName(pszExpectedDriverName)) |
212 | 0 | { |
213 | 0 | return true; |
214 | 0 | } |
215 | | |
216 | 33.5k | if ((!poOpenInfo->papszAllowedDrivers || |
217 | 33.5k | CSLFindString(poOpenInfo->papszAllowedDrivers, "JSONFG") >= 0) && |
218 | 33.5k | GDALGetDriverByName("JSONFG") && JSONFGIsObject(pszText, poOpenInfo)) |
219 | 1.26k | { |
220 | 1.26k | return false; |
221 | 1.26k | } |
222 | | |
223 | 32.2k | if (osTopLevelType == "FeatureCollection") |
224 | 3.07k | { |
225 | 3.07k | return true; |
226 | 3.07k | } |
227 | | |
228 | 29.2k | const std::string osWithoutSpace = GetCompactJSon(pszText, strlen(pszText)); |
229 | 29.2k | if (osWithoutSpace.find("{\"features\":[") == 0 && |
230 | 29.2k | osWithoutSpace.find(szESRIJSonFeaturesGeometryRings) != 0 && |
231 | 29.2k | osWithoutSpace.find(szESRIJSonFeaturesAttributes) != 0) |
232 | 7.22k | { |
233 | 7.22k | return true; |
234 | 7.22k | } |
235 | | |
236 | | // See |
237 | | // https://raw.githubusercontent.com/icepack/icepack-data/master/meshes/larsen/larsen_inflow.geojson |
238 | | // "{"crs":...,"features":[..." |
239 | | // or |
240 | | // https://gist.githubusercontent.com/NiklasDallmann/27e339dd78d1942d524fbcd179f9fdcf/raw/527a8319d75a9e29446a32a19e4c902213b0d668/42XR9nLAh8Poh9Xmniqh3Bs9iisNm74mYMC56v3Wfyo=_isochrones_fails.geojson |
241 | | // "{"bbox":...,"features":[..." |
242 | 21.9k | if (osWithoutSpace.find(",\"features\":[") != std::string::npos) |
243 | 3.04k | { |
244 | 3.04k | return !ESRIJSONIsObject(pszText, poOpenInfo); |
245 | 3.04k | } |
246 | | |
247 | | // See https://github.com/OSGeo/gdal/issues/2720 |
248 | 18.9k | if (osWithoutSpace.find("{\"coordinates\":[") == 0 || |
249 | | // and https://github.com/OSGeo/gdal/issues/2787 |
250 | 18.9k | osWithoutSpace.find("{\"geometry\":{\"coordinates\":[") == 0 || |
251 | | // and https://github.com/qgis/QGIS/issues/61266 |
252 | 18.9k | osWithoutSpace.find( |
253 | 18.5k | "{\"geometry\":{\"type\":\"Point\",\"coordinates\":[") == 0 || |
254 | 18.9k | osWithoutSpace.find( |
255 | 18.5k | "{\"geometry\":{\"type\":\"LineString\",\"coordinates\":[") == 0 || |
256 | 18.9k | osWithoutSpace.find( |
257 | 18.4k | "{\"geometry\":{\"type\":\"Polygon\",\"coordinates\":[") == 0 || |
258 | 18.9k | osWithoutSpace.find( |
259 | 18.4k | "{\"geometry\":{\"type\":\"MultiPoint\",\"coordinates\":[") == 0 || |
260 | 18.9k | osWithoutSpace.find( |
261 | 18.4k | "{\"geometry\":{\"type\":\"MultiLineString\",\"coordinates\":[") == |
262 | 18.4k | 0 || |
263 | 18.9k | osWithoutSpace.find( |
264 | 18.4k | "{\"geometry\":{\"type\":\"MultiPolygon\",\"coordinates\":[") == |
265 | 18.4k | 0 || |
266 | 18.9k | osWithoutSpace.find("{\"geometry\":{\"type\":\"GeometryCollection\"," |
267 | 18.4k | "\"geometries\":[") == 0) |
268 | 500 | { |
269 | 500 | return true; |
270 | 500 | } |
271 | | |
272 | 18.4k | if (osTopLevelType == "Feature" || osTopLevelType == "Point" || |
273 | 18.4k | osTopLevelType == "LineString" || osTopLevelType == "Polygon" || |
274 | 18.4k | osTopLevelType == "MultiPoint" || osTopLevelType == "MultiLineString" || |
275 | 18.4k | osTopLevelType == "MultiPolygon" || |
276 | 18.4k | osTopLevelType == "GeometryCollection") |
277 | 7.94k | { |
278 | 7.94k | bMightBeSequence = true; |
279 | 7.94k | return true; |
280 | 7.94k | } |
281 | | |
282 | | // See https://github.com/OSGeo/gdal/issues/3280 |
283 | 10.4k | if (osWithoutSpace.find("{\"properties\":{") == 0) |
284 | 3.89k | { |
285 | 3.89k | bMightBeSequence = true; |
286 | 3.89k | bReadMoreBytes = true; |
287 | 3.89k | return false; |
288 | 3.89k | } |
289 | | |
290 | 6.59k | return false; |
291 | 10.4k | } |
292 | | |
293 | | static bool IsGeoJSONLikeObject(const char *pszText, GDALOpenInfo *poOpenInfo, |
294 | | const char *pszExpectedDriverName) |
295 | 4.07k | { |
296 | 4.07k | bool bMightBeSequence; |
297 | 4.07k | bool bReadMoreBytes; |
298 | 4.07k | return IsGeoJSONLikeObject(pszText, bMightBeSequence, bReadMoreBytes, |
299 | 4.07k | poOpenInfo, pszExpectedDriverName); |
300 | 4.07k | } |
301 | | |
302 | | /************************************************************************/ |
303 | | /* ESRIJSONIsObject() */ |
304 | | /************************************************************************/ |
305 | | |
306 | | bool ESRIJSONIsObject(const char *pszText, GDALOpenInfo *poOpenInfo) |
307 | 131k | { |
308 | 131k | if (!IsJSONObject(pszText)) |
309 | 120k | return false; |
310 | | |
311 | 11.2k | if (poOpenInfo->IsSingleAllowedDriver("ESRIJSON") && |
312 | 11.2k | GDALGetDriverByName("ESRIJSON")) |
313 | 0 | { |
314 | 0 | return true; |
315 | 0 | } |
316 | | |
317 | 11.2k | if ( // ESRI Json geometry |
318 | 11.2k | (strstr(pszText, "\"geometryType\"") != nullptr && |
319 | 11.2k | strstr(pszText, "\"esriGeometry") != nullptr) |
320 | | |
321 | | // ESRI Json "FeatureCollection" |
322 | 11.2k | || strstr(pszText, "\"fieldAliases\"") != nullptr |
323 | | |
324 | | // ESRI Json "FeatureCollection" |
325 | 11.2k | || (strstr(pszText, "\"fields\"") != nullptr && |
326 | 10.5k | strstr(pszText, "\"esriFieldType") != nullptr)) |
327 | 759 | { |
328 | 759 | return true; |
329 | 759 | } |
330 | | |
331 | 10.5k | const std::string osWithoutSpace = GetCompactJSon(pszText, strlen(pszText)); |
332 | 10.5k | if (osWithoutSpace.find(szESRIJSonFeaturesGeometryRings) == 0 || |
333 | 10.5k | osWithoutSpace.find(szESRIJSonFeaturesAttributes) == 0 || |
334 | 10.5k | osWithoutSpace.find("\"spatialReference\":{\"wkid\":") != |
335 | 10.5k | std::string::npos) |
336 | 357 | { |
337 | 357 | return true; |
338 | 357 | } |
339 | | |
340 | 10.1k | return false; |
341 | 10.5k | } |
342 | | |
343 | | /************************************************************************/ |
344 | | /* TopoJSONIsObject() */ |
345 | | /************************************************************************/ |
346 | | |
347 | | bool TopoJSONIsObject(const char *pszText, GDALOpenInfo *poOpenInfo) |
348 | 128k | { |
349 | 128k | if (!IsJSONObject(pszText)) |
350 | 120k | return false; |
351 | | |
352 | 7.72k | if (poOpenInfo->IsSingleAllowedDriver("TopoJSON") && |
353 | 7.72k | GDALGetDriverByName("TopoJSON")) |
354 | 0 | { |
355 | 0 | return true; |
356 | 0 | } |
357 | | |
358 | 7.72k | return GetTopLevelType(pszText) == "Topology"; |
359 | 7.72k | } |
360 | | |
361 | | /************************************************************************/ |
362 | | /* IsLikelyNewlineSequenceGeoJSON() */ |
363 | | /************************************************************************/ |
364 | | |
365 | | static GDALIdentifyEnum |
366 | | IsLikelyNewlineSequenceGeoJSON(VSILFILE *fpL, const GByte *pabyHeader, |
367 | | const char *pszFileContent) |
368 | 7.41k | { |
369 | 7.41k | const size_t nBufferSize = 4096 * 10; |
370 | 7.41k | std::vector<GByte> abyBuffer; |
371 | 7.41k | abyBuffer.resize(nBufferSize + 1); |
372 | | |
373 | 7.41k | int nCurlLevel = 0; |
374 | 7.41k | bool bInString = false; |
375 | 7.41k | bool bLastIsEscape = false; |
376 | 7.41k | bool bFirstIter = true; |
377 | 7.41k | bool bEOLFound = false; |
378 | 7.41k | int nCountObject = 0; |
379 | 12.5k | while (true) |
380 | 12.5k | { |
381 | 12.5k | size_t nRead; |
382 | 12.5k | bool bEnd = false; |
383 | 12.5k | if (bFirstIter) |
384 | 7.41k | { |
385 | 7.41k | const char *pszText = |
386 | 7.41k | pszFileContent ? pszFileContent |
387 | 7.41k | : reinterpret_cast<const char *>(pabyHeader); |
388 | 7.41k | assert(pszText); |
389 | 7.41k | nRead = std::min(strlen(pszText), nBufferSize); |
390 | 7.41k | memcpy(abyBuffer.data(), pszText, nRead); |
391 | 7.41k | bFirstIter = false; |
392 | 7.41k | if (fpL) |
393 | 7.00k | { |
394 | 7.00k | VSIFSeekL(fpL, nRead, SEEK_SET); |
395 | 7.00k | } |
396 | 7.41k | } |
397 | 5.08k | else |
398 | 5.08k | { |
399 | 5.08k | nRead = VSIFReadL(abyBuffer.data(), 1, nBufferSize, fpL); |
400 | 5.08k | bEnd = nRead < nBufferSize; |
401 | 5.08k | } |
402 | 75.3M | for (size_t i = 0; i < nRead; i++) |
403 | 75.3M | { |
404 | 75.3M | if (nCurlLevel == 0) |
405 | 22.1k | { |
406 | 22.1k | if (abyBuffer[i] == '{') |
407 | 12.4k | { |
408 | 12.4k | nCountObject++; |
409 | 12.4k | if (nCountObject == 2) |
410 | 5.05k | { |
411 | 5.05k | break; |
412 | 5.05k | } |
413 | 7.40k | nCurlLevel++; |
414 | 7.40k | } |
415 | 9.67k | else if (nCountObject == 1 && abyBuffer[i] == '\n') |
416 | 5.45k | { |
417 | 5.45k | bEOLFound = true; |
418 | 5.45k | } |
419 | 4.21k | else if (!isspace(static_cast<unsigned char>(abyBuffer[i]))) |
420 | 707 | { |
421 | 707 | return GDAL_IDENTIFY_FALSE; |
422 | 707 | } |
423 | 22.1k | } |
424 | 75.3M | else if (bInString) |
425 | 35.6M | { |
426 | 35.6M | if (bLastIsEscape) |
427 | 52.6k | { |
428 | 52.6k | bLastIsEscape = false; |
429 | 52.6k | } |
430 | 35.5M | else if (abyBuffer[i] == '\\') |
431 | 52.6k | { |
432 | 52.6k | bLastIsEscape = true; |
433 | 52.6k | } |
434 | 35.5M | else if (abyBuffer[i] == '"') |
435 | 481k | { |
436 | 481k | bInString = false; |
437 | 481k | } |
438 | 35.6M | } |
439 | 39.7M | else if (abyBuffer[i] == '"') |
440 | 482k | { |
441 | 482k | bInString = true; |
442 | 482k | } |
443 | 39.2M | else if (abyBuffer[i] == '{') |
444 | 238k | { |
445 | 238k | nCurlLevel++; |
446 | 238k | } |
447 | 39.0M | else if (abyBuffer[i] == '}') |
448 | 36.9k | { |
449 | 36.9k | nCurlLevel--; |
450 | 36.9k | } |
451 | 75.3M | } |
452 | 11.7k | if (!fpL || bEnd || nCountObject == 2) |
453 | 6.71k | break; |
454 | 11.7k | } |
455 | 6.71k | if (bEOLFound && nCountObject == 2) |
456 | 5.03k | return GDAL_IDENTIFY_TRUE; |
457 | 1.68k | return GDAL_IDENTIFY_UNKNOWN; |
458 | 6.71k | } |
459 | | |
460 | | /************************************************************************/ |
461 | | /* GeoJSONFileIsObject() */ |
462 | | /************************************************************************/ |
463 | | |
464 | | static bool GeoJSONFileIsObject(GDALOpenInfo *poOpenInfo) |
465 | 135k | { |
466 | | // By default read first 6000 bytes. |
467 | | // 6000 was chosen as enough bytes to |
468 | | // enable all current tests to pass. |
469 | | |
470 | 135k | if (poOpenInfo->fpL == nullptr || !poOpenInfo->TryToIngest(6000)) |
471 | 46.5k | { |
472 | 46.5k | return false; |
473 | 46.5k | } |
474 | | |
475 | 88.6k | bool bMightBeSequence = false; |
476 | 88.6k | bool bReadMoreBytes = false; |
477 | 88.6k | if (!IsGeoJSONLikeObject( |
478 | 88.6k | reinterpret_cast<const char *>(poOpenInfo->pabyHeader), |
479 | 88.6k | bMightBeSequence, bReadMoreBytes, poOpenInfo, "GeoJSON")) |
480 | 80.1k | { |
481 | 80.1k | if (!(bReadMoreBytes && poOpenInfo->nHeaderBytes >= 6000 && |
482 | 80.1k | poOpenInfo->TryToIngest(1000 * 1000) && |
483 | 80.1k | !IsGeoJSONLikeObject( |
484 | 922 | reinterpret_cast<const char *>(poOpenInfo->pabyHeader), |
485 | 922 | bMightBeSequence, bReadMoreBytes, poOpenInfo, "GeoJSON"))) |
486 | 79.2k | { |
487 | 79.2k | return false; |
488 | 79.2k | } |
489 | 80.1k | } |
490 | | |
491 | 9.32k | return !(bMightBeSequence && IsLikelyNewlineSequenceGeoJSON( |
492 | 3.39k | poOpenInfo->fpL, poOpenInfo->pabyHeader, |
493 | 3.39k | nullptr) == GDAL_IDENTIFY_TRUE); |
494 | 88.6k | } |
495 | | |
496 | | /************************************************************************/ |
497 | | /* GeoJSONIsObject() */ |
498 | | /************************************************************************/ |
499 | | |
500 | | bool GeoJSONIsObject(const char *pszText, GDALOpenInfo *poOpenInfo) |
501 | 139k | { |
502 | 139k | bool bMightBeSequence = false; |
503 | 139k | bool bReadMoreBytes = false; |
504 | 139k | if (!IsGeoJSONLikeObject(pszText, bMightBeSequence, bReadMoreBytes, |
505 | 139k | poOpenInfo, "GeoJSON")) |
506 | 135k | { |
507 | 135k | return false; |
508 | 135k | } |
509 | | |
510 | 3.37k | return !(bMightBeSequence && |
511 | 3.37k | IsLikelyNewlineSequenceGeoJSON(nullptr, nullptr, pszText) == |
512 | 414 | GDAL_IDENTIFY_TRUE); |
513 | 139k | } |
514 | | |
515 | | /************************************************************************/ |
516 | | /* GeoJSONSeqFileIsObject() */ |
517 | | /************************************************************************/ |
518 | | |
519 | | static bool GeoJSONSeqFileIsObject(GDALOpenInfo *poOpenInfo) |
520 | 133k | { |
521 | | // By default read first 6000 bytes. |
522 | | // 6000 was chosen as enough bytes to |
523 | | // enable all current tests to pass. |
524 | | |
525 | 133k | if (poOpenInfo->fpL == nullptr || !poOpenInfo->TryToIngest(6000)) |
526 | 46.5k | { |
527 | 46.5k | return false; |
528 | 46.5k | } |
529 | | |
530 | 87.4k | const char *pszText = |
531 | 87.4k | reinterpret_cast<const char *>(poOpenInfo->pabyHeader); |
532 | 87.4k | if (pszText[0] == '\x1e') |
533 | 4.07k | return IsGeoJSONLikeObject(pszText + 1, poOpenInfo, "GeoJSONSeq"); |
534 | | |
535 | 83.3k | bool bMightBeSequence = false; |
536 | 83.3k | bool bReadMoreBytes = false; |
537 | 83.3k | if (!IsGeoJSONLikeObject(pszText, bMightBeSequence, bReadMoreBytes, |
538 | 83.3k | poOpenInfo, "GeoJSONSeq")) |
539 | 77.2k | { |
540 | 77.2k | if (!(bReadMoreBytes && poOpenInfo->nHeaderBytes >= 6000 && |
541 | 77.2k | poOpenInfo->TryToIngest(1000 * 1000) && |
542 | 77.2k | IsGeoJSONLikeObject( |
543 | 457 | reinterpret_cast<const char *>(poOpenInfo->pabyHeader), |
544 | 457 | bMightBeSequence, bReadMoreBytes, poOpenInfo, "GeoJSONSeq"))) |
545 | 77.2k | { |
546 | 77.2k | return false; |
547 | 77.2k | } |
548 | 77.2k | } |
549 | | |
550 | 6.11k | if (poOpenInfo->IsSingleAllowedDriver("GeoJSONSeq") && |
551 | 6.11k | IsLikelyNewlineSequenceGeoJSON(poOpenInfo->fpL, poOpenInfo->pabyHeader, |
552 | 0 | nullptr) != GDAL_IDENTIFY_FALSE && |
553 | 6.11k | GDALGetDriverByName("GeoJSONSeq")) |
554 | 0 | { |
555 | 0 | return true; |
556 | 0 | } |
557 | | |
558 | 6.11k | return bMightBeSequence && IsLikelyNewlineSequenceGeoJSON( |
559 | 3.60k | poOpenInfo->fpL, poOpenInfo->pabyHeader, |
560 | 3.60k | nullptr) == GDAL_IDENTIFY_TRUE; |
561 | 6.11k | } |
562 | | |
563 | | bool GeoJSONSeqIsObject(const char *pszText, GDALOpenInfo *poOpenInfo) |
564 | 133k | { |
565 | 133k | if (pszText[0] == '\x1e') |
566 | 1 | return IsGeoJSONLikeObject(pszText + 1, poOpenInfo, "GeoJSONSeq"); |
567 | | |
568 | 133k | bool bMightBeSequence = false; |
569 | 133k | bool bReadMoreBytes = false; |
570 | 133k | if (!IsGeoJSONLikeObject(pszText, bMightBeSequence, bReadMoreBytes, |
571 | 133k | poOpenInfo, "GeoJSONSeq")) |
572 | 133k | { |
573 | 133k | return false; |
574 | 133k | } |
575 | | |
576 | 0 | if (poOpenInfo->IsSingleAllowedDriver("GeoJSONSeq") && |
577 | 0 | IsLikelyNewlineSequenceGeoJSON(nullptr, nullptr, pszText) != |
578 | 0 | GDAL_IDENTIFY_FALSE && |
579 | 0 | GDALGetDriverByName("GeoJSONSeq")) |
580 | 0 | { |
581 | 0 | return true; |
582 | 0 | } |
583 | | |
584 | 0 | return bMightBeSequence && |
585 | 0 | IsLikelyNewlineSequenceGeoJSON(nullptr, nullptr, pszText) == |
586 | 0 | GDAL_IDENTIFY_TRUE; |
587 | 0 | } |
588 | | |
589 | | /************************************************************************/ |
590 | | /* JSONFGFileIsObject() */ |
591 | | /************************************************************************/ |
592 | | |
593 | | static bool JSONFGFileIsObject(GDALOpenInfo *poOpenInfo) |
594 | 91.4k | { |
595 | | // 6000 somewhat arbitrary. Based on other JSON-like drivers |
596 | 91.4k | if (poOpenInfo->fpL == nullptr || !poOpenInfo->TryToIngest(6000)) |
597 | 45.7k | { |
598 | 45.7k | return false; |
599 | 45.7k | } |
600 | | |
601 | 45.6k | const char *pszText = |
602 | 45.6k | reinterpret_cast<const char *>(poOpenInfo->pabyHeader); |
603 | 45.6k | return JSONFGIsObject(pszText, poOpenInfo); |
604 | 91.4k | } |
605 | | |
606 | | bool JSONFGIsObject(const char *pszText, GDALOpenInfo *poOpenInfo) |
607 | 171k | { |
608 | 171k | if (!IsJSONObject(pszText)) |
609 | 131k | return false; |
610 | | |
611 | 39.6k | if (poOpenInfo->IsSingleAllowedDriver("JSONFG") && |
612 | 39.6k | GDALGetDriverByName("JSONFG")) |
613 | 0 | { |
614 | 0 | return true; |
615 | 0 | } |
616 | | |
617 | 39.6k | const std::string osWithoutSpace = GetCompactJSon(pszText, strlen(pszText)); |
618 | | |
619 | | // In theory, conformsTo should be required, but let be lax... |
620 | 39.6k | { |
621 | 39.6k | const auto nPos = osWithoutSpace.find("\"conformsTo\":["); |
622 | 39.6k | if (nPos != std::string::npos) |
623 | 1.78k | { |
624 | 1.78k | for (const char *pszVersion : {"0.1", "0.2", "0.3"}) |
625 | 5.23k | { |
626 | 5.23k | if (osWithoutSpace.find( |
627 | 5.23k | CPLSPrintf("\"[ogc-json-fg-1-%s:core]\"", pszVersion), |
628 | 5.23k | nPos) != std::string::npos || |
629 | 5.23k | osWithoutSpace.find( |
630 | 5.08k | CPLSPrintf( |
631 | 5.08k | "\"http://www.opengis.net/spec/json-fg-1/%s\"", |
632 | 5.08k | pszVersion), |
633 | 5.08k | nPos) != std::string::npos) |
634 | 184 | { |
635 | 184 | return true; |
636 | 184 | } |
637 | 5.23k | } |
638 | 1.78k | } |
639 | 39.6k | } |
640 | | |
641 | 39.4k | if (osWithoutSpace.find("\"place\":{\"type\":") != std::string::npos || |
642 | 39.4k | osWithoutSpace.find("\"place\":{\"coordinates\":") != |
643 | 39.0k | std::string::npos || |
644 | 39.4k | osWithoutSpace.find("\"time\":{\"date\":") != std::string::npos || |
645 | 39.4k | osWithoutSpace.find("\"time\":{\"timestamp\":") != std::string::npos || |
646 | 39.4k | osWithoutSpace.find("\"time\":{\"interval\":") != std::string::npos) |
647 | 1.46k | { |
648 | 1.46k | return true; |
649 | 1.46k | } |
650 | | |
651 | 37.9k | if (osWithoutSpace.find("\"coordRefSys\":") != std::string::npos || |
652 | 37.9k | osWithoutSpace.find("\"featureType\":") != std::string::npos) |
653 | 2.99k | { |
654 | | // Check that coordRefSys and/or featureType are either at the |
655 | | // FeatureCollection or Feature level |
656 | 2.99k | struct MyParser : public CPLJSonStreamingParser |
657 | 2.99k | { |
658 | 2.99k | bool m_bFoundJSONFGFeatureType = false; |
659 | 2.99k | bool m_bFoundJSONFGCoordrefSys = false; |
660 | 2.99k | std::string m_osLevel{}; |
661 | | |
662 | 2.99k | void StartObjectMember(const char *pszKey, size_t nLength) override |
663 | 40.4k | { |
664 | 40.4k | if (nLength == strlen("featureType") && |
665 | 40.4k | strcmp(pszKey, "featureType") == 0) |
666 | 637 | { |
667 | 637 | m_bFoundJSONFGFeatureType = |
668 | 637 | (m_osLevel == "{" || // At FeatureCollection level |
669 | 637 | m_osLevel == "{[{"); // At Feature level |
670 | 637 | if (m_bFoundJSONFGFeatureType) |
671 | 250 | StopParsing(); |
672 | 637 | } |
673 | 39.8k | else if (nLength == strlen("coordRefSys") && |
674 | 39.8k | strcmp(pszKey, "coordRefSys") == 0) |
675 | 1.74k | { |
676 | 1.74k | m_bFoundJSONFGCoordrefSys = |
677 | 1.74k | (m_osLevel == "{" || // At FeatureCollection level |
678 | 1.74k | m_osLevel == "{[{"); // At Feature level |
679 | 1.74k | if (m_bFoundJSONFGCoordrefSys) |
680 | 494 | StopParsing(); |
681 | 1.74k | } |
682 | 40.4k | } |
683 | | |
684 | 2.99k | void StartObject() override |
685 | 18.0k | { |
686 | 18.0k | m_osLevel += '{'; |
687 | 18.0k | } |
688 | | |
689 | 2.99k | void EndObject() override |
690 | 3.27k | { |
691 | 3.27k | if (!m_osLevel.empty()) |
692 | 3.27k | m_osLevel.pop_back(); |
693 | 3.27k | } |
694 | | |
695 | 2.99k | void StartArray() override |
696 | 14.9k | { |
697 | 14.9k | m_osLevel += '['; |
698 | 14.9k | } |
699 | | |
700 | 2.99k | void EndArray() override |
701 | 8.13k | { |
702 | 8.13k | if (!m_osLevel.empty()) |
703 | 8.13k | m_osLevel.pop_back(); |
704 | 8.13k | } |
705 | 2.99k | }; |
706 | | |
707 | 2.99k | MyParser oParser; |
708 | 2.99k | oParser.Parse(pszText, strlen(pszText), true); |
709 | 2.99k | if (oParser.m_bFoundJSONFGFeatureType || |
710 | 2.99k | oParser.m_bFoundJSONFGCoordrefSys) |
711 | 744 | { |
712 | 744 | return true; |
713 | 744 | } |
714 | 2.99k | } |
715 | | |
716 | 37.2k | return false; |
717 | 37.9k | } |
718 | | |
719 | | /************************************************************************/ |
720 | | /* IsLikelyESRIJSONURL() */ |
721 | | /************************************************************************/ |
722 | | |
723 | | static bool IsLikelyESRIJSONURL(const char *pszURL) |
724 | 5.41k | { |
725 | | // URLs with f=json are strong candidates for ESRI JSON services |
726 | | // except if they have "/items?", in which case they are likely OAPIF |
727 | 5.41k | return (strstr(pszURL, "f=json") != nullptr || |
728 | 5.41k | strstr(pszURL, "f=pjson") != nullptr || |
729 | 5.41k | strstr(pszURL, "resultRecordCount=") != nullptr) && |
730 | 5.41k | strstr(pszURL, "/items?") == nullptr; |
731 | 5.41k | } |
732 | | |
733 | | /************************************************************************/ |
734 | | /* GeoJSONGetSourceType() */ |
735 | | /************************************************************************/ |
736 | | |
737 | | GeoJSONSourceType GeoJSONGetSourceType(GDALOpenInfo *poOpenInfo) |
738 | 136k | { |
739 | 136k | GeoJSONSourceType srcType = eGeoJSONSourceUnknown; |
740 | | |
741 | | // NOTE: Sometimes URL ends with .geojson token, for example |
742 | | // http://example/path/2232.geojson |
743 | | // It's important to test beginning of source first. |
744 | 136k | if (STARTS_WITH_CI(poOpenInfo->pszFilename, "GEOJSON:http://") || |
745 | 136k | STARTS_WITH_CI(poOpenInfo->pszFilename, "GEOJSON:https://") || |
746 | 136k | STARTS_WITH_CI(poOpenInfo->pszFilename, "GEOJSON:ftp://")) |
747 | 2 | { |
748 | 2 | srcType = eGeoJSONSourceService; |
749 | 2 | } |
750 | 136k | else if (STARTS_WITH_CI(poOpenInfo->pszFilename, "http://") || |
751 | 136k | STARTS_WITH_CI(poOpenInfo->pszFilename, "https://") || |
752 | 136k | STARTS_WITH_CI(poOpenInfo->pszFilename, "ftp://")) |
753 | 1.16k | { |
754 | 1.16k | if (poOpenInfo->IsSingleAllowedDriver("GeoJSON")) |
755 | 0 | { |
756 | 0 | return eGeoJSONSourceService; |
757 | 0 | } |
758 | 1.16k | if ((strstr(poOpenInfo->pszFilename, "SERVICE=WFS") || |
759 | 1.16k | strstr(poOpenInfo->pszFilename, "service=WFS") || |
760 | 1.16k | strstr(poOpenInfo->pszFilename, "service=wfs")) && |
761 | 1.16k | !strstr(poOpenInfo->pszFilename, "json")) |
762 | 59 | { |
763 | 59 | return eGeoJSONSourceUnknown; |
764 | 59 | } |
765 | 1.10k | if (IsLikelyESRIJSONURL(poOpenInfo->pszFilename)) |
766 | 29 | { |
767 | 29 | return eGeoJSONSourceUnknown; |
768 | 29 | } |
769 | 1.07k | srcType = eGeoJSONSourceService; |
770 | 1.07k | } |
771 | 135k | else if (STARTS_WITH_CI(poOpenInfo->pszFilename, "GeoJSON:")) |
772 | 1 | { |
773 | 1 | VSIStatBufL sStat; |
774 | 1 | if (VSIStatL(poOpenInfo->pszFilename + strlen("GeoJSON:"), &sStat) == 0) |
775 | 0 | { |
776 | 0 | return eGeoJSONSourceFile; |
777 | 0 | } |
778 | 1 | const char *pszText = poOpenInfo->pszFilename + strlen("GeoJSON:"); |
779 | 1 | if (GeoJSONIsObject(pszText, poOpenInfo)) |
780 | 0 | return eGeoJSONSourceText; |
781 | 1 | return eGeoJSONSourceUnknown; |
782 | 1 | } |
783 | 135k | else if (GeoJSONIsObject(poOpenInfo->pszFilename, poOpenInfo)) |
784 | 0 | { |
785 | 0 | srcType = eGeoJSONSourceText; |
786 | 0 | } |
787 | 135k | else if (GeoJSONFileIsObject(poOpenInfo)) |
788 | 7.64k | { |
789 | 7.64k | srcType = eGeoJSONSourceFile; |
790 | 7.64k | } |
791 | | |
792 | 136k | return srcType; |
793 | 136k | } |
794 | | |
795 | | /************************************************************************/ |
796 | | /* ESRIJSONDriverGetSourceType() */ |
797 | | /************************************************************************/ |
798 | | |
799 | | GeoJSONSourceType ESRIJSONDriverGetSourceType(GDALOpenInfo *poOpenInfo) |
800 | 129k | { |
801 | 129k | if (STARTS_WITH_CI(poOpenInfo->pszFilename, "ESRIJSON:http://") || |
802 | 129k | STARTS_WITH_CI(poOpenInfo->pszFilename, "ESRIJSON:https://") || |
803 | 129k | STARTS_WITH_CI(poOpenInfo->pszFilename, "ESRIJSON:ftp://")) |
804 | 6 | { |
805 | 6 | return eGeoJSONSourceService; |
806 | 6 | } |
807 | 129k | else if (STARTS_WITH(poOpenInfo->pszFilename, "http://") || |
808 | 129k | STARTS_WITH(poOpenInfo->pszFilename, "https://") || |
809 | 129k | STARTS_WITH(poOpenInfo->pszFilename, "ftp://")) |
810 | 653 | { |
811 | 653 | if (poOpenInfo->IsSingleAllowedDriver("ESRIJSON")) |
812 | 0 | { |
813 | 0 | return eGeoJSONSourceService; |
814 | 0 | } |
815 | 653 | if (IsLikelyESRIJSONURL(poOpenInfo->pszFilename)) |
816 | 58 | { |
817 | 58 | return eGeoJSONSourceService; |
818 | 58 | } |
819 | 595 | return eGeoJSONSourceUnknown; |
820 | 653 | } |
821 | | |
822 | 128k | if (STARTS_WITH_CI(poOpenInfo->pszFilename, "ESRIJSON:")) |
823 | 32 | { |
824 | 32 | VSIStatBufL sStat; |
825 | 32 | if (VSIStatL(poOpenInfo->pszFilename + strlen("ESRIJSON:"), &sStat) == |
826 | 32 | 0) |
827 | 0 | { |
828 | 0 | return eGeoJSONSourceFile; |
829 | 0 | } |
830 | 32 | const char *pszText = poOpenInfo->pszFilename + strlen("ESRIJSON:"); |
831 | 32 | if (ESRIJSONIsObject(pszText, poOpenInfo)) |
832 | 0 | return eGeoJSONSourceText; |
833 | 32 | return eGeoJSONSourceUnknown; |
834 | 32 | } |
835 | | |
836 | 128k | if (poOpenInfo->fpL == nullptr) |
837 | 46.5k | { |
838 | 46.5k | const char *pszText = poOpenInfo->pszFilename; |
839 | 46.5k | if (ESRIJSONIsObject(pszText, poOpenInfo)) |
840 | 70 | return eGeoJSONSourceText; |
841 | 46.5k | return eGeoJSONSourceUnknown; |
842 | 46.5k | } |
843 | | |
844 | | // By default read first 6000 bytes. |
845 | | // 6000 was chosen as enough bytes to |
846 | | // enable all current tests to pass. |
847 | 82.0k | if (!poOpenInfo->TryToIngest(6000)) |
848 | 0 | { |
849 | 0 | return eGeoJSONSourceUnknown; |
850 | 0 | } |
851 | | |
852 | 82.0k | if (poOpenInfo->pabyHeader != nullptr && |
853 | 82.0k | ESRIJSONIsObject(reinterpret_cast<const char *>(poOpenInfo->pabyHeader), |
854 | 82.0k | poOpenInfo)) |
855 | 934 | { |
856 | 934 | return eGeoJSONSourceFile; |
857 | 934 | } |
858 | 81.1k | return eGeoJSONSourceUnknown; |
859 | 82.0k | } |
860 | | |
861 | | /************************************************************************/ |
862 | | /* TopoJSONDriverGetSourceType() */ |
863 | | /************************************************************************/ |
864 | | |
865 | | GeoJSONSourceType TopoJSONDriverGetSourceType(GDALOpenInfo *poOpenInfo) |
866 | 129k | { |
867 | 129k | if (STARTS_WITH_CI(poOpenInfo->pszFilename, "TopoJSON:http://") || |
868 | 129k | STARTS_WITH_CI(poOpenInfo->pszFilename, "TopoJSON:https://") || |
869 | 129k | STARTS_WITH_CI(poOpenInfo->pszFilename, "TopoJSON:ftp://")) |
870 | 16 | { |
871 | 16 | return eGeoJSONSourceService; |
872 | 16 | } |
873 | 129k | else if (STARTS_WITH(poOpenInfo->pszFilename, "http://") || |
874 | 129k | STARTS_WITH(poOpenInfo->pszFilename, "https://") || |
875 | 129k | STARTS_WITH(poOpenInfo->pszFilename, "ftp://")) |
876 | 1.21k | { |
877 | 1.21k | if (poOpenInfo->IsSingleAllowedDriver("TOPOJSON")) |
878 | 0 | { |
879 | 0 | return eGeoJSONSourceService; |
880 | 0 | } |
881 | 1.21k | if (IsLikelyESRIJSONURL(poOpenInfo->pszFilename)) |
882 | 29 | { |
883 | 29 | return eGeoJSONSourceUnknown; |
884 | 29 | } |
885 | 1.19k | return eGeoJSONSourceService; |
886 | 1.21k | } |
887 | | |
888 | 128k | if (STARTS_WITH_CI(poOpenInfo->pszFilename, "TopoJSON:")) |
889 | 90 | { |
890 | 90 | VSIStatBufL sStat; |
891 | 90 | if (VSIStatL(poOpenInfo->pszFilename + strlen("TopoJSON:"), &sStat) == |
892 | 90 | 0) |
893 | 0 | { |
894 | 0 | return eGeoJSONSourceFile; |
895 | 0 | } |
896 | 90 | const char *pszText = poOpenInfo->pszFilename + strlen("TopoJSON:"); |
897 | 90 | if (TopoJSONIsObject(pszText, poOpenInfo)) |
898 | 0 | return eGeoJSONSourceText; |
899 | 90 | return eGeoJSONSourceUnknown; |
900 | 90 | } |
901 | | |
902 | 128k | if (poOpenInfo->fpL == nullptr) |
903 | 46.4k | { |
904 | 46.4k | const char *pszText = poOpenInfo->pszFilename; |
905 | 46.4k | if (TopoJSONIsObject(pszText, poOpenInfo)) |
906 | 0 | return eGeoJSONSourceText; |
907 | 46.4k | return eGeoJSONSourceUnknown; |
908 | 46.4k | } |
909 | | |
910 | | // By default read first 6000 bytes. |
911 | | // 6000 was chosen as enough bytes to |
912 | | // enable all current tests to pass. |
913 | 81.5k | if (!poOpenInfo->TryToIngest(6000)) |
914 | 0 | { |
915 | 0 | return eGeoJSONSourceUnknown; |
916 | 0 | } |
917 | | |
918 | 81.5k | if (poOpenInfo->pabyHeader != nullptr && |
919 | 81.5k | TopoJSONIsObject(reinterpret_cast<const char *>(poOpenInfo->pabyHeader), |
920 | 81.5k | poOpenInfo)) |
921 | 68 | { |
922 | 68 | return eGeoJSONSourceFile; |
923 | 68 | } |
924 | 81.5k | return eGeoJSONSourceUnknown; |
925 | 81.5k | } |
926 | | |
927 | | /************************************************************************/ |
928 | | /* GeoJSONSeqGetSourceType() */ |
929 | | /************************************************************************/ |
930 | | |
931 | | GeoJSONSourceType GeoJSONSeqGetSourceType(GDALOpenInfo *poOpenInfo) |
932 | 135k | { |
933 | 135k | GeoJSONSourceType srcType = eGeoJSONSourceUnknown; |
934 | | |
935 | 135k | if (STARTS_WITH_CI(poOpenInfo->pszFilename, "GEOJSONSeq:http://") || |
936 | 135k | STARTS_WITH_CI(poOpenInfo->pszFilename, "GEOJSONSeq:https://") || |
937 | 135k | STARTS_WITH_CI(poOpenInfo->pszFilename, "GEOJSONSeq:ftp://")) |
938 | 0 | { |
939 | 0 | srcType = eGeoJSONSourceService; |
940 | 0 | } |
941 | 135k | else if (STARTS_WITH_CI(poOpenInfo->pszFilename, "http://") || |
942 | 135k | STARTS_WITH_CI(poOpenInfo->pszFilename, "https://") || |
943 | 135k | STARTS_WITH_CI(poOpenInfo->pszFilename, "ftp://")) |
944 | 1.21k | { |
945 | 1.21k | if (poOpenInfo->IsSingleAllowedDriver("GeoJSONSeq")) |
946 | 0 | { |
947 | 0 | return eGeoJSONSourceService; |
948 | 0 | } |
949 | 1.21k | if (IsLikelyESRIJSONURL(poOpenInfo->pszFilename)) |
950 | 29 | { |
951 | 29 | return eGeoJSONSourceUnknown; |
952 | 29 | } |
953 | 1.19k | srcType = eGeoJSONSourceService; |
954 | 1.19k | } |
955 | 134k | else if (STARTS_WITH_CI(poOpenInfo->pszFilename, "GEOJSONSeq:")) |
956 | 158 | { |
957 | 158 | VSIStatBufL sStat; |
958 | 158 | if (VSIStatL(poOpenInfo->pszFilename + strlen("GEOJSONSeq:"), &sStat) == |
959 | 158 | 0) |
960 | 140 | { |
961 | 140 | return eGeoJSONSourceFile; |
962 | 140 | } |
963 | 18 | const char *pszText = poOpenInfo->pszFilename + strlen("GEOJSONSeq:"); |
964 | 18 | if (GeoJSONSeqIsObject(pszText, poOpenInfo)) |
965 | 0 | return eGeoJSONSourceText; |
966 | 18 | return eGeoJSONSourceUnknown; |
967 | 18 | } |
968 | 133k | else if (GeoJSONSeqIsObject(poOpenInfo->pszFilename, poOpenInfo)) |
969 | 0 | { |
970 | 0 | srcType = eGeoJSONSourceText; |
971 | 0 | } |
972 | 133k | else if (GeoJSONSeqFileIsObject(poOpenInfo)) |
973 | 7.11k | { |
974 | 7.11k | srcType = eGeoJSONSourceFile; |
975 | 7.11k | } |
976 | | |
977 | 135k | return srcType; |
978 | 135k | } |
979 | | |
980 | | /************************************************************************/ |
981 | | /* JSONFGDriverGetSourceType() */ |
982 | | /************************************************************************/ |
983 | | |
984 | | GeoJSONSourceType JSONFGDriverGetSourceType(GDALOpenInfo *poOpenInfo) |
985 | 93.2k | { |
986 | 93.2k | GeoJSONSourceType srcType = eGeoJSONSourceUnknown; |
987 | | |
988 | 93.2k | if (STARTS_WITH_CI(poOpenInfo->pszFilename, "JSONFG:http://") || |
989 | 93.2k | STARTS_WITH_CI(poOpenInfo->pszFilename, "JSONFG:https://") || |
990 | 93.2k | STARTS_WITH_CI(poOpenInfo->pszFilename, "JSONFG:ftp://")) |
991 | 108 | { |
992 | 108 | srcType = eGeoJSONSourceService; |
993 | 108 | } |
994 | 93.1k | else if (STARTS_WITH_CI(poOpenInfo->pszFilename, "http://") || |
995 | 93.1k | STARTS_WITH_CI(poOpenInfo->pszFilename, "https://") || |
996 | 93.1k | STARTS_WITH_CI(poOpenInfo->pszFilename, "ftp://")) |
997 | 1.21k | { |
998 | 1.21k | if (poOpenInfo->IsSingleAllowedDriver("JSONFG")) |
999 | 0 | { |
1000 | 0 | return eGeoJSONSourceService; |
1001 | 0 | } |
1002 | 1.21k | if (IsLikelyESRIJSONURL(poOpenInfo->pszFilename)) |
1003 | 29 | { |
1004 | 29 | return eGeoJSONSourceUnknown; |
1005 | 29 | } |
1006 | 1.19k | srcType = eGeoJSONSourceService; |
1007 | 1.19k | } |
1008 | 91.9k | else if (STARTS_WITH_CI(poOpenInfo->pszFilename, "JSONFG:")) |
1009 | 304 | { |
1010 | 304 | VSIStatBufL sStat; |
1011 | 304 | const size_t nJSONFGPrefixLen = strlen("JSONFG:"); |
1012 | 304 | if (VSIStatL(poOpenInfo->pszFilename + nJSONFGPrefixLen, &sStat) == 0) |
1013 | 100 | { |
1014 | 100 | return eGeoJSONSourceFile; |
1015 | 100 | } |
1016 | 204 | const char *pszText = poOpenInfo->pszFilename + nJSONFGPrefixLen; |
1017 | 204 | if (JSONFGIsObject(pszText, poOpenInfo)) |
1018 | 20 | return eGeoJSONSourceText; |
1019 | 184 | return eGeoJSONSourceUnknown; |
1020 | 204 | } |
1021 | 91.6k | else if (JSONFGIsObject(poOpenInfo->pszFilename, poOpenInfo)) |
1022 | 256 | { |
1023 | 256 | srcType = eGeoJSONSourceText; |
1024 | 256 | } |
1025 | 91.4k | else if (JSONFGFileIsObject(poOpenInfo)) |
1026 | 850 | { |
1027 | 850 | srcType = eGeoJSONSourceFile; |
1028 | 850 | } |
1029 | | |
1030 | 92.9k | return srcType; |
1031 | 93.2k | } |
1032 | | |
1033 | | /************************************************************************/ |
1034 | | /* GeoJSONStringPropertyToFieldType() */ |
1035 | | /************************************************************************/ |
1036 | | |
1037 | | OGRFieldType GeoJSONStringPropertyToFieldType(json_object *poObject, |
1038 | | int &nTZFlag) |
1039 | 32.3k | { |
1040 | 32.3k | if (poObject == nullptr) |
1041 | 1.04k | { |
1042 | 1.04k | return OFTString; |
1043 | 1.04k | } |
1044 | 31.3k | const char *pszStr = json_object_get_string(poObject); |
1045 | | |
1046 | 31.3k | nTZFlag = 0; |
1047 | 31.3k | OGRField sWrkField; |
1048 | 31.3k | CPLPushErrorHandler(CPLQuietErrorHandler); |
1049 | 31.3k | const bool bSuccess = CPL_TO_BOOL(OGRParseDate(pszStr, &sWrkField, 0)); |
1050 | 31.3k | CPLPopErrorHandler(); |
1051 | 31.3k | CPLErrorReset(); |
1052 | 31.3k | if (bSuccess) |
1053 | 817 | { |
1054 | 817 | const bool bHasDate = |
1055 | 817 | strchr(pszStr, '/') != nullptr || strchr(pszStr, '-') != nullptr; |
1056 | 817 | const bool bHasTime = strchr(pszStr, ':') != nullptr; |
1057 | 817 | nTZFlag = sWrkField.Date.TZFlag; |
1058 | 817 | if (bHasDate && bHasTime) |
1059 | 817 | return OFTDateTime; |
1060 | 0 | else if (bHasDate) |
1061 | 0 | return OFTDate; |
1062 | 0 | else |
1063 | 0 | return OFTTime; |
1064 | | // TODO: What if both are false? |
1065 | 817 | } |
1066 | 30.4k | return OFTString; |
1067 | 31.3k | } |
1068 | | |
1069 | | /************************************************************************/ |
1070 | | /* GeoJSONHTTPFetchWithContentTypeHeader() */ |
1071 | | /************************************************************************/ |
1072 | | |
1073 | | CPLHTTPResult *GeoJSONHTTPFetchWithContentTypeHeader(const char *pszURL) |
1074 | 1.76k | { |
1075 | 1.76k | std::string osHeaders; |
1076 | 1.76k | const char *pszGDAL_HTTP_HEADERS = |
1077 | 1.76k | CPLGetConfigOption("GDAL_HTTP_HEADERS", nullptr); |
1078 | 1.76k | bool bFoundAcceptHeader = false; |
1079 | 1.76k | if (pszGDAL_HTTP_HEADERS) |
1080 | 0 | { |
1081 | 0 | bool bHeadersDone = false; |
1082 | | // Compatibility hack for "HEADERS=Accept: text/plain, application/json" |
1083 | 0 | if (strstr(pszGDAL_HTTP_HEADERS, "\r\n") == nullptr) |
1084 | 0 | { |
1085 | 0 | const char *pszComma = strchr(pszGDAL_HTTP_HEADERS, ','); |
1086 | 0 | if (pszComma != nullptr && strchr(pszComma, ':') == nullptr) |
1087 | 0 | { |
1088 | 0 | osHeaders = pszGDAL_HTTP_HEADERS; |
1089 | 0 | bFoundAcceptHeader = |
1090 | 0 | STARTS_WITH_CI(pszGDAL_HTTP_HEADERS, "Accept:"); |
1091 | 0 | bHeadersDone = true; |
1092 | 0 | } |
1093 | 0 | } |
1094 | 0 | if (!bHeadersDone) |
1095 | 0 | { |
1096 | | // We accept both raw headers with \r\n as a separator, or as |
1097 | | // a comma separated list of foo: bar values. |
1098 | 0 | const CPLStringList aosTokens( |
1099 | 0 | strstr(pszGDAL_HTTP_HEADERS, "\r\n") |
1100 | 0 | ? CSLTokenizeString2(pszGDAL_HTTP_HEADERS, "\r\n", 0) |
1101 | 0 | : CSLTokenizeString2(pszGDAL_HTTP_HEADERS, ",", |
1102 | 0 | CSLT_HONOURSTRINGS)); |
1103 | 0 | for (int i = 0; i < aosTokens.size(); ++i) |
1104 | 0 | { |
1105 | 0 | if (!osHeaders.empty()) |
1106 | 0 | osHeaders += "\r\n"; |
1107 | 0 | if (!bFoundAcceptHeader) |
1108 | 0 | bFoundAcceptHeader = |
1109 | 0 | STARTS_WITH_CI(aosTokens[i], "Accept:"); |
1110 | 0 | osHeaders += aosTokens[i]; |
1111 | 0 | } |
1112 | 0 | } |
1113 | 0 | } |
1114 | 1.76k | if (!bFoundAcceptHeader) |
1115 | 1.76k | { |
1116 | 1.76k | if (!osHeaders.empty()) |
1117 | 0 | osHeaders += "\r\n"; |
1118 | 1.76k | osHeaders += "Accept: text/plain, application/json"; |
1119 | 1.76k | } |
1120 | | |
1121 | 1.76k | CPLStringList aosOptions; |
1122 | 1.76k | aosOptions.SetNameValue("HEADERS", osHeaders.c_str()); |
1123 | 1.76k | CPLHTTPResult *pResult = CPLHTTPFetch(pszURL, aosOptions.List()); |
1124 | | |
1125 | 1.76k | if (nullptr == pResult || 0 == pResult->nDataLen || |
1126 | 1.76k | 0 != CPLGetLastErrorNo()) |
1127 | 1.76k | { |
1128 | 1.76k | CPLHTTPDestroyResult(pResult); |
1129 | 1.76k | return nullptr; |
1130 | 1.76k | } |
1131 | | |
1132 | 0 | if (0 != pResult->nStatus) |
1133 | 0 | { |
1134 | 0 | CPLError(CE_Failure, CPLE_AppDefined, "Curl reports error: %d: %s", |
1135 | 0 | pResult->nStatus, pResult->pszErrBuf); |
1136 | 0 | CPLHTTPDestroyResult(pResult); |
1137 | 0 | return nullptr; |
1138 | 0 | } |
1139 | | |
1140 | 0 | return pResult; |
1141 | 0 | } |