/src/cpython3/Objects/stringlib/repr.h
Line | Count | Source |
1 | | /* stringlib: repr() implementation */ |
2 | | |
3 | | #ifndef STRINGLIB_FASTSEARCH_H |
4 | | #error must include "stringlib/fastsearch.h" before including this module |
5 | | #endif |
6 | | |
7 | | |
8 | | static void |
9 | | STRINGLIB(repr)(PyObject *unicode, Py_UCS4 quote, |
10 | | STRINGLIB_CHAR *odata) |
11 | 1.53k | { |
12 | 1.53k | Py_ssize_t isize = PyUnicode_GET_LENGTH(unicode); |
13 | 1.53k | const void *idata = PyUnicode_DATA(unicode); |
14 | 1.53k | int ikind = PyUnicode_KIND(unicode); |
15 | | |
16 | 0 | *odata++ = quote; |
17 | 4.19M | for (Py_ssize_t i = 0; i < isize; i++) { |
18 | 4.19M | Py_UCS4 ch = PyUnicode_READ(ikind, idata, i); |
19 | | |
20 | | /* Escape quotes and backslashes */ |
21 | 4.19M | if ((ch == quote) || (ch == '\\')) { |
22 | 27.5k | *odata++ = '\\'; |
23 | 27.5k | *odata++ = ch; |
24 | 27.5k | continue; |
25 | 27.5k | } |
26 | | |
27 | | /* Map special whitespace to '\t', \n', '\r' */ |
28 | 4.16M | if (ch == '\t') { |
29 | 4.67k | *odata++ = '\\'; |
30 | 4.67k | *odata++ = 't'; |
31 | 4.67k | } |
32 | 4.16M | else if (ch == '\n') { |
33 | 11.7k | *odata++ = '\\'; |
34 | 11.7k | *odata++ = 'n'; |
35 | 11.7k | } |
36 | 4.15M | else if (ch == '\r') { |
37 | 3.40k | *odata++ = '\\'; |
38 | 3.40k | *odata++ = 'r'; |
39 | 3.40k | } |
40 | | |
41 | | /* Map non-printable US ASCII to '\xhh' */ |
42 | 4.14M | else if (ch < ' ' || ch == 0x7F) { |
43 | 1.63M | *odata++ = '\\'; |
44 | 1.63M | *odata++ = 'x'; |
45 | 1.63M | *odata++ = Py_hexdigits[(ch >> 4) & 0x000F]; |
46 | 1.63M | *odata++ = Py_hexdigits[ch & 0x000F]; |
47 | 1.63M | } |
48 | | |
49 | | /* Copy ASCII characters as-is */ |
50 | 2.51M | else if (ch < 0x7F) { |
51 | 2.04M | *odata++ = ch; |
52 | 2.04M | } |
53 | | |
54 | | /* Non-ASCII characters */ |
55 | 464k | else { |
56 | | /* Map Unicode whitespace and control characters |
57 | | (categories Z* and C* except ASCII space) |
58 | | */ |
59 | 464k | if (!Py_UNICODE_ISPRINTABLE(ch)) { |
60 | 313k | *odata++ = '\\'; |
61 | | /* Map 8-bit characters to '\xhh' */ |
62 | 313k | if (ch <= 0xff) { |
63 | 310k | *odata++ = 'x'; |
64 | 310k | *odata++ = Py_hexdigits[(ch >> 4) & 0x000F]; |
65 | 310k | *odata++ = Py_hexdigits[ch & 0x000F]; |
66 | 310k | } |
67 | | /* Map 16-bit characters to '\uxxxx' */ |
68 | 3.21k | else if (ch <= 0xffff) { |
69 | 2.50k | *odata++ = 'u'; |
70 | 2.50k | *odata++ = Py_hexdigits[(ch >> 12) & 0xF]; |
71 | 2.50k | *odata++ = Py_hexdigits[(ch >> 8) & 0xF]; |
72 | 2.50k | *odata++ = Py_hexdigits[(ch >> 4) & 0xF]; |
73 | 2.50k | *odata++ = Py_hexdigits[ch & 0xF]; |
74 | 2.50k | } |
75 | | /* Map 21-bit characters to '\U00xxxxxx' */ |
76 | 715 | else { |
77 | 715 | *odata++ = 'U'; |
78 | 715 | *odata++ = Py_hexdigits[(ch >> 28) & 0xF]; |
79 | 715 | *odata++ = Py_hexdigits[(ch >> 24) & 0xF]; |
80 | 715 | *odata++ = Py_hexdigits[(ch >> 20) & 0xF]; |
81 | 715 | *odata++ = Py_hexdigits[(ch >> 16) & 0xF]; |
82 | 715 | *odata++ = Py_hexdigits[(ch >> 12) & 0xF]; |
83 | 715 | *odata++ = Py_hexdigits[(ch >> 8) & 0xF]; |
84 | 715 | *odata++ = Py_hexdigits[(ch >> 4) & 0xF]; |
85 | 715 | *odata++ = Py_hexdigits[ch & 0xF]; |
86 | 715 | } |
87 | 313k | } |
88 | | /* Copy characters as-is */ |
89 | 150k | else { |
90 | 150k | *odata++ = ch; |
91 | 150k | } |
92 | 464k | } |
93 | 4.16M | } |
94 | 1.53k | *odata = quote; |
95 | 1.53k | } unicodeobject.c:ucs1lib_repr Line | Count | Source | 11 | 1.13k | { | 12 | 1.13k | Py_ssize_t isize = PyUnicode_GET_LENGTH(unicode); | 13 | 1.13k | const void *idata = PyUnicode_DATA(unicode); | 14 | 1.13k | int ikind = PyUnicode_KIND(unicode); | 15 | | | 16 | 0 | *odata++ = quote; | 17 | 2.20M | for (Py_ssize_t i = 0; i < isize; i++) { | 18 | 2.19M | Py_UCS4 ch = PyUnicode_READ(ikind, idata, i); | 19 | | | 20 | | /* Escape quotes and backslashes */ | 21 | 2.19M | if ((ch == quote) || (ch == '\\')) { | 22 | 15.2k | *odata++ = '\\'; | 23 | 15.2k | *odata++ = ch; | 24 | 15.2k | continue; | 25 | 15.2k | } | 26 | | | 27 | | /* Map special whitespace to '\t', \n', '\r' */ | 28 | 2.18M | if (ch == '\t') { | 29 | 2.86k | *odata++ = '\\'; | 30 | 2.86k | *odata++ = 't'; | 31 | 2.86k | } | 32 | 2.18M | else if (ch == '\n') { | 33 | 3.98k | *odata++ = '\\'; | 34 | 3.98k | *odata++ = 'n'; | 35 | 3.98k | } | 36 | 2.17M | else if (ch == '\r') { | 37 | 2.03k | *odata++ = '\\'; | 38 | 2.03k | *odata++ = 'r'; | 39 | 2.03k | } | 40 | | | 41 | | /* Map non-printable US ASCII to '\xhh' */ | 42 | 2.17M | else if (ch < ' ' || ch == 0x7F) { | 43 | 773k | *odata++ = '\\'; | 44 | 773k | *odata++ = 'x'; | 45 | 773k | *odata++ = Py_hexdigits[(ch >> 4) & 0x000F]; | 46 | 773k | *odata++ = Py_hexdigits[ch & 0x000F]; | 47 | 773k | } | 48 | | | 49 | | /* Copy ASCII characters as-is */ | 50 | 1.40M | else if (ch < 0x7F) { | 51 | 960k | *odata++ = ch; | 52 | 960k | } | 53 | | | 54 | | /* Non-ASCII characters */ | 55 | 441k | else { | 56 | | /* Map Unicode whitespace and control characters | 57 | | (categories Z* and C* except ASCII space) | 58 | | */ | 59 | 441k | if (!Py_UNICODE_ISPRINTABLE(ch)) { | 60 | 308k | *odata++ = '\\'; | 61 | | /* Map 8-bit characters to '\xhh' */ | 62 | 308k | if (ch <= 0xff) { | 63 | 306k | *odata++ = 'x'; | 64 | 306k | *odata++ = Py_hexdigits[(ch >> 4) & 0x000F]; | 65 | 306k | *odata++ = Py_hexdigits[ch & 0x000F]; | 66 | 306k | } | 67 | | /* Map 16-bit characters to '\uxxxx' */ | 68 | 1.47k | else if (ch <= 0xffff) { | 69 | 1.12k | *odata++ = 'u'; | 70 | 1.12k | *odata++ = Py_hexdigits[(ch >> 12) & 0xF]; | 71 | 1.12k | *odata++ = Py_hexdigits[(ch >> 8) & 0xF]; | 72 | 1.12k | *odata++ = Py_hexdigits[(ch >> 4) & 0xF]; | 73 | 1.12k | *odata++ = Py_hexdigits[ch & 0xF]; | 74 | 1.12k | } | 75 | | /* Map 21-bit characters to '\U00xxxxxx' */ | 76 | 355 | else { | 77 | 355 | *odata++ = 'U'; | 78 | 355 | *odata++ = Py_hexdigits[(ch >> 28) & 0xF]; | 79 | 355 | *odata++ = Py_hexdigits[(ch >> 24) & 0xF]; | 80 | 355 | *odata++ = Py_hexdigits[(ch >> 20) & 0xF]; | 81 | 355 | *odata++ = Py_hexdigits[(ch >> 16) & 0xF]; | 82 | 355 | *odata++ = Py_hexdigits[(ch >> 12) & 0xF]; | 83 | 355 | *odata++ = Py_hexdigits[(ch >> 8) & 0xF]; | 84 | 355 | *odata++ = Py_hexdigits[(ch >> 4) & 0xF]; | 85 | 355 | *odata++ = Py_hexdigits[ch & 0xF]; | 86 | 355 | } | 87 | 308k | } | 88 | | /* Copy characters as-is */ | 89 | 133k | else { | 90 | 133k | *odata++ = ch; | 91 | 133k | } | 92 | 441k | } | 93 | 2.18M | } | 94 | 1.13k | *odata = quote; | 95 | 1.13k | } |
unicodeobject.c:ucs2lib_repr Line | Count | Source | 11 | 215 | { | 12 | 215 | Py_ssize_t isize = PyUnicode_GET_LENGTH(unicode); | 13 | 215 | const void *idata = PyUnicode_DATA(unicode); | 14 | 215 | int ikind = PyUnicode_KIND(unicode); | 15 | | | 16 | 0 | *odata++ = quote; | 17 | 928k | for (Py_ssize_t i = 0; i < isize; i++) { | 18 | 927k | Py_UCS4 ch = PyUnicode_READ(ikind, idata, i); | 19 | | | 20 | | /* Escape quotes and backslashes */ | 21 | 927k | if ((ch == quote) || (ch == '\\')) { | 22 | 7.22k | *odata++ = '\\'; | 23 | 7.22k | *odata++ = ch; | 24 | 7.22k | continue; | 25 | 7.22k | } | 26 | | | 27 | | /* Map special whitespace to '\t', \n', '\r' */ | 28 | 920k | if (ch == '\t') { | 29 | 1.48k | *odata++ = '\\'; | 30 | 1.48k | *odata++ = 't'; | 31 | 1.48k | } | 32 | 919k | else if (ch == '\n') { | 33 | 2.74k | *odata++ = '\\'; | 34 | 2.74k | *odata++ = 'n'; | 35 | 2.74k | } | 36 | 916k | else if (ch == '\r') { | 37 | 596 | *odata++ = '\\'; | 38 | 596 | *odata++ = 'r'; | 39 | 596 | } | 40 | | | 41 | | /* Map non-printable US ASCII to '\xhh' */ | 42 | 915k | else if (ch < ' ' || ch == 0x7F) { | 43 | 264k | *odata++ = '\\'; | 44 | 264k | *odata++ = 'x'; | 45 | 264k | *odata++ = Py_hexdigits[(ch >> 4) & 0x000F]; | 46 | 264k | *odata++ = Py_hexdigits[ch & 0x000F]; | 47 | 264k | } | 48 | | | 49 | | /* Copy ASCII characters as-is */ | 50 | 651k | else if (ch < 0x7F) { | 51 | 632k | *odata++ = ch; | 52 | 632k | } | 53 | | | 54 | | /* Non-ASCII characters */ | 55 | 19.0k | else { | 56 | | /* Map Unicode whitespace and control characters | 57 | | (categories Z* and C* except ASCII space) | 58 | | */ | 59 | 19.0k | if (!Py_UNICODE_ISPRINTABLE(ch)) { | 60 | 4.41k | *odata++ = '\\'; | 61 | | /* Map 8-bit characters to '\xhh' */ | 62 | 4.41k | if (ch <= 0xff) { | 63 | 3.14k | *odata++ = 'x'; | 64 | 3.14k | *odata++ = Py_hexdigits[(ch >> 4) & 0x000F]; | 65 | 3.14k | *odata++ = Py_hexdigits[ch & 0x000F]; | 66 | 3.14k | } | 67 | | /* Map 16-bit characters to '\uxxxx' */ | 68 | 1.27k | else if (ch <= 0xffff) { | 69 | 1.02k | *odata++ = 'u'; | 70 | 1.02k | *odata++ = Py_hexdigits[(ch >> 12) & 0xF]; | 71 | 1.02k | *odata++ = Py_hexdigits[(ch >> 8) & 0xF]; | 72 | 1.02k | *odata++ = Py_hexdigits[(ch >> 4) & 0xF]; | 73 | 1.02k | *odata++ = Py_hexdigits[ch & 0xF]; | 74 | 1.02k | } | 75 | | /* Map 21-bit characters to '\U00xxxxxx' */ | 76 | 247 | else { | 77 | 247 | *odata++ = 'U'; | 78 | 247 | *odata++ = Py_hexdigits[(ch >> 28) & 0xF]; | 79 | 247 | *odata++ = Py_hexdigits[(ch >> 24) & 0xF]; | 80 | 247 | *odata++ = Py_hexdigits[(ch >> 20) & 0xF]; | 81 | 247 | *odata++ = Py_hexdigits[(ch >> 16) & 0xF]; | 82 | 247 | *odata++ = Py_hexdigits[(ch >> 12) & 0xF]; | 83 | 247 | *odata++ = Py_hexdigits[(ch >> 8) & 0xF]; | 84 | 247 | *odata++ = Py_hexdigits[(ch >> 4) & 0xF]; | 85 | 247 | *odata++ = Py_hexdigits[ch & 0xF]; | 86 | 247 | } | 87 | 4.41k | } | 88 | | /* Copy characters as-is */ | 89 | 14.6k | else { | 90 | 14.6k | *odata++ = ch; | 91 | 14.6k | } | 92 | 19.0k | } | 93 | 920k | } | 94 | 215 | *odata = quote; | 95 | 215 | } |
unicodeobject.c:ucs4lib_repr Line | Count | Source | 11 | 187 | { | 12 | 187 | Py_ssize_t isize = PyUnicode_GET_LENGTH(unicode); | 13 | 187 | const void *idata = PyUnicode_DATA(unicode); | 14 | 187 | int ikind = PyUnicode_KIND(unicode); | 15 | | | 16 | 0 | *odata++ = quote; | 17 | 1.06M | for (Py_ssize_t i = 0; i < isize; i++) { | 18 | 1.06M | Py_UCS4 ch = PyUnicode_READ(ikind, idata, i); | 19 | | | 20 | | /* Escape quotes and backslashes */ | 21 | 1.06M | if ((ch == quote) || (ch == '\\')) { | 22 | 5.04k | *odata++ = '\\'; | 23 | 5.04k | *odata++ = ch; | 24 | 5.04k | continue; | 25 | 5.04k | } | 26 | | | 27 | | /* Map special whitespace to '\t', \n', '\r' */ | 28 | 1.06M | if (ch == '\t') { | 29 | 329 | *odata++ = '\\'; | 30 | 329 | *odata++ = 't'; | 31 | 329 | } | 32 | 1.06M | else if (ch == '\n') { | 33 | 5.03k | *odata++ = '\\'; | 34 | 5.03k | *odata++ = 'n'; | 35 | 5.03k | } | 36 | 1.05M | else if (ch == '\r') { | 37 | 773 | *odata++ = '\\'; | 38 | 773 | *odata++ = 'r'; | 39 | 773 | } | 40 | | | 41 | | /* Map non-printable US ASCII to '\xhh' */ | 42 | 1.05M | else if (ch < ' ' || ch == 0x7F) { | 43 | 598k | *odata++ = '\\'; | 44 | 598k | *odata++ = 'x'; | 45 | 598k | *odata++ = Py_hexdigits[(ch >> 4) & 0x000F]; | 46 | 598k | *odata++ = Py_hexdigits[ch & 0x000F]; | 47 | 598k | } | 48 | | | 49 | | /* Copy ASCII characters as-is */ | 50 | 460k | else if (ch < 0x7F) { | 51 | 456k | *odata++ = ch; | 52 | 456k | } | 53 | | | 54 | | /* Non-ASCII characters */ | 55 | 3.24k | else { | 56 | | /* Map Unicode whitespace and control characters | 57 | | (categories Z* and C* except ASCII space) | 58 | | */ | 59 | 3.24k | if (!Py_UNICODE_ISPRINTABLE(ch)) { | 60 | 1.40k | *odata++ = '\\'; | 61 | | /* Map 8-bit characters to '\xhh' */ | 62 | 1.40k | if (ch <= 0xff) { | 63 | 933 | *odata++ = 'x'; | 64 | 933 | *odata++ = Py_hexdigits[(ch >> 4) & 0x000F]; | 65 | 933 | *odata++ = Py_hexdigits[ch & 0x000F]; | 66 | 933 | } | 67 | | /* Map 16-bit characters to '\uxxxx' */ | 68 | 471 | else if (ch <= 0xffff) { | 69 | 358 | *odata++ = 'u'; | 70 | 358 | *odata++ = Py_hexdigits[(ch >> 12) & 0xF]; | 71 | 358 | *odata++ = Py_hexdigits[(ch >> 8) & 0xF]; | 72 | 358 | *odata++ = Py_hexdigits[(ch >> 4) & 0xF]; | 73 | 358 | *odata++ = Py_hexdigits[ch & 0xF]; | 74 | 358 | } | 75 | | /* Map 21-bit characters to '\U00xxxxxx' */ | 76 | 113 | else { | 77 | 113 | *odata++ = 'U'; | 78 | 113 | *odata++ = Py_hexdigits[(ch >> 28) & 0xF]; | 79 | 113 | *odata++ = Py_hexdigits[(ch >> 24) & 0xF]; | 80 | 113 | *odata++ = Py_hexdigits[(ch >> 20) & 0xF]; | 81 | 113 | *odata++ = Py_hexdigits[(ch >> 16) & 0xF]; | 82 | 113 | *odata++ = Py_hexdigits[(ch >> 12) & 0xF]; | 83 | 113 | *odata++ = Py_hexdigits[(ch >> 8) & 0xF]; | 84 | 113 | *odata++ = Py_hexdigits[(ch >> 4) & 0xF]; | 85 | 113 | *odata++ = Py_hexdigits[ch & 0xF]; | 86 | 113 | } | 87 | 1.40k | } | 88 | | /* Copy characters as-is */ | 89 | 1.84k | else { | 90 | 1.84k | *odata++ = ch; | 91 | 1.84k | } | 92 | 3.24k | } | 93 | 1.06M | } | 94 | 187 | *odata = quote; | 95 | 187 | } |
|