summaryrefslogtreecommitdiffstats
path: root/Python/suggestions.c
blob: c336ec8ffffc9c4151ed34c21774594d55b2a678 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
#include "Python.h"
#include "pycore_frame.h"

#include "pycore_pyerrors.h"
#include "pycore_code.h"        // _PyCode_GetVarnames()

#define MAX_CANDIDATE_ITEMS 750
#define MAX_STRING_SIZE 40

#define MOVE_COST 2
#define CASE_COST 1

#define LEAST_FIVE_BITS(n) ((n) & 31)

static inline int
substitution_cost(char a, char b)
{
    if (LEAST_FIVE_BITS(a) != LEAST_FIVE_BITS(b)) {
        // Not the same, not a case flip.
        return MOVE_COST;
    }
    if (a == b) {
        return 0;
    }
    if ('A' <= a && a <= 'Z') {
        a += ('a' - 'A');
    }
    if ('A' <= b && b <= 'Z') {
        b += ('a' - 'A');
    }
    if (a == b) {
        return CASE_COST;
    }
    return MOVE_COST;
}

/* Calculate the Levenshtein distance between string1 and string2 */
static Py_ssize_t
levenshtein_distance(const char *a, size_t a_size,
                     const char *b, size_t b_size,
                     size_t max_cost)
{
    static size_t buffer[MAX_STRING_SIZE];

    // Both strings are the same (by identity)
    if (a == b) {
        return 0;
    }

    // Trim away common affixes.
    while (a_size && b_size && a[0] == b[0]) {
        a++; a_size--;
        b++; b_size--;
    }
    while (a_size && b_size && a[a_size-1] == b[b_size-1]) {
        a_size--;
        b_size--;
    }
    if (a_size == 0 || b_size == 0) {
        return (a_size + b_size) * MOVE_COST;
    }
    if (a_size > MAX_STRING_SIZE || b_size > MAX_STRING_SIZE) {
        return max_cost + 1;
    }

    // Prefer shorter buffer
    if (b_size < a_size) {
        const char *t = a; a = b; b = t;
        size_t t_size = a_size; a_size = b_size; b_size = t_size;
    }

    // quick fail when a match is impossible.
    if ((b_size - a_size) * MOVE_COST > max_cost) {
        return max_cost + 1;
    }

    // Instead of producing the whole traditional len(a)-by-len(b)
    // matrix, we can update just one row in place.
    // Initialize the buffer row
    size_t tmp = MOVE_COST;
    for (size_t i = 0; i < a_size; i++) {
        // cost from b[:0] to a[:i+1]
        buffer[i] = tmp;
        tmp += MOVE_COST;
    }

    size_t result = 0;
    for (size_t b_index = 0; b_index < b_size; b_index++) {
        char code = b[b_index];
        // cost(b[:b_index], a[:0]) == b_index * MOVE_COST
        size_t distance = result = b_index * MOVE_COST;
        size_t minimum = SIZE_MAX;
        for (size_t index = 0; index < a_size; index++) {

            // cost(b[:b_index+1], a[:index+1]) = min(
            //     // 1) substitute
            //     cost(b[:b_index], a[:index])
            //         + substitution_cost(b[b_index], a[index]),
            //     // 2) delete from b
            //     cost(b[:b_index], a[:index+1]) + MOVE_COST,
            //     // 3) delete from a
            //     cost(b[:b_index+1], a[index]) + MOVE_COST
            // )

            // 1) Previous distance in this row is cost(b[:b_index], a[:index])
            size_t substitute = distance + substitution_cost(code, a[index]);
            // 2) cost(b[:b_index], a[:index+1]) from previous row
            distance = buffer[index];
            // 3) existing result is cost(b[:b_index+1], a[index])

            size_t insert_delete = Py_MIN(result, distance) + MOVE_COST;
            result = Py_MIN(insert_delete, substitute);

            // cost(b[:b_index+1], a[:index+1])
            buffer[index] = result;
            if (result < minimum) {
                minimum = result;
            }
        }
        if (minimum > max_cost) {
            // Everything in this row is too big, so bail early.
            return max_cost + 1;
        }
    }
    return result;
}

static inline PyObject *
calculate_suggestions(PyObject *dir,
                      PyObject *name)
{
    assert(!PyErr_Occurred());
    assert(PyList_CheckExact(dir));

    Py_ssize_t dir_size = PyList_GET_SIZE(dir);
    if (dir_size >= MAX_CANDIDATE_ITEMS) {
        return NULL;
    }

    Py_ssize_t suggestion_distance = PY_SSIZE_T_MAX;
    PyObject *suggestion = NULL;
    Py_ssize_t name_size;
    const char *name_str = PyUnicode_AsUTF8AndSize(name, &name_size);
    if (name_str == NULL) {
        return NULL;
    }

    for (int i = 0; i < dir_size; ++i) {
        PyObject *item = PyList_GET_ITEM(dir, i);
        Py_ssize_t item_size;
        const char *item_str = PyUnicode_AsUTF8AndSize(item, &item_size);
        if (item_str == NULL) {
            return NULL;
        }
        if (PyUnicode_CompareWithASCIIString(name, item_str) == 0) {
            continue;
        }
        // No more than 1/3 of the involved characters should need changed.
        Py_ssize_t max_distance = (name_size + item_size + 3) * MOVE_COST / 6;
        // Don't take matches we've already beaten.
        max_distance = Py_MIN(max_distance, suggestion_distance - 1);
        Py_ssize_t current_distance =
            levenshtein_distance(name_str, name_size,
                                 item_str, item_size, max_distance);
        if (current_distance > max_distance) {
            continue;
        }
        if (!suggestion || current_distance < suggestion_distance) {
            suggestion = item;
            suggestion_distance = current_distance;
        }
    }
    Py_XINCREF(suggestion);
    return suggestion;
}

static PyObject *
offer_suggestions_for_attribute_error(PyAttributeErrorObject *exc)
{
    PyObject *name = exc->name; // borrowed reference
    PyObject *obj = exc->obj; // borrowed reference

    // Abort if we don't have an attribute name or we have an invalid one
    if (name == NULL || obj == NULL || !PyUnicode_CheckExact(name)) {
        return NULL;
    }

    PyObject *dir = PyObject_Dir(obj);
    if (dir == NULL) {
        return NULL;
    }

    PyObject *suggestions = calculate_suggestions(dir, name);
    Py_DECREF(dir);
    return suggestions;
}


static PyObject *
offer_suggestions_for_name_error(PyNameErrorObject *exc)
{
    PyObject *name = exc->name; // borrowed reference
    PyTracebackObject *traceback = (PyTracebackObject *) exc->traceback; // borrowed reference
    // Abort if we don't have a variable name or we have an invalid one
    // or if we don't have a traceback to work with
    if (name == NULL || !PyUnicode_CheckExact(name) ||
        traceback == NULL || !Py_IS_TYPE(traceback, &PyTraceBack_Type)
    ) {
        return NULL;
    }

    // Move to the traceback of the exception
    while (1) {
        PyTracebackObject *next = traceback->tb_next;
        if (next == NULL || !Py_IS_TYPE(next, &PyTraceBack_Type)) {
            break;
        }
        else {
            traceback = next;
        }
    }

    PyFrameObject *frame = traceback->tb_frame;
    assert(frame != NULL);
    PyCodeObject *code = PyFrame_GetCode(frame);
    assert(code != NULL && code->co_localsplusnames != NULL);
    PyObject *varnames = _PyCode_GetVarnames(code);
    if (varnames == NULL) {
        return NULL;
    }
    PyObject *dir = PySequence_List(varnames);
    Py_DECREF(varnames);
    Py_DECREF(code);
    if (dir == NULL) {
        return NULL;
    }

    PyObject *suggestions = calculate_suggestions(dir, name);
    Py_DECREF(dir);
    if (suggestions != NULL) {
        return suggestions;
    }

    dir = PySequence_List(frame->f_frame->f_globals);
    if (dir == NULL) {
        return NULL;
    }
    suggestions = calculate_suggestions(dir, name);
    Py_DECREF(dir);
    if (suggestions != NULL) {
        return suggestions;
    }

    dir = PySequence_List(frame->f_frame->f_builtins);
    if (dir == NULL) {
        return NULL;
    }
    suggestions = calculate_suggestions(dir, name);
    Py_DECREF(dir);

    return suggestions;
}

// Offer suggestions for a given exception. Returns a python string object containing the
// suggestions. This function returns NULL if no suggestion was found or if an exception happened,
// users must call PyErr_Occurred() to disambiguate.
PyObject *
_Py_Offer_Suggestions(PyObject *exception)
{
    PyObject *result = NULL;
    assert(!PyErr_Occurred());
    if (Py_IS_TYPE(exception, (PyTypeObject*)PyExc_AttributeError)) {
        result = offer_suggestions_for_attribute_error((PyAttributeErrorObject *) exception);
    } else if (Py_IS_TYPE(exception, (PyTypeObject*)PyExc_NameError)) {
        result = offer_suggestions_for_name_error((PyNameErrorObject *) exception);
    }
    return result;
}

Py_ssize_t
_Py_UTF8_Edit_Cost(PyObject *a, PyObject *b, Py_ssize_t max_cost)
{
    assert(PyUnicode_Check(a) && PyUnicode_Check(b));
    Py_ssize_t size_a, size_b;
    const char *utf8_a = PyUnicode_AsUTF8AndSize(a, &size_a);
    if (utf8_a == NULL) {
        return -1;
    }
    const char *utf8_b = PyUnicode_AsUTF8AndSize(b, &size_b);
    if (utf8_b == NULL) {
        return -1;
    }
    if (max_cost == -1) {
        max_cost = MOVE_COST * Py_MAX(size_a, size_b);
    }
    return levenshtein_distance(utf8_a, size_a, utf8_b, size_b, max_cost);
}