Compiled pandas with -Wextra (#56327)

* Compiled pandas with -Wextra

* np_datetime_strings.c fallthrough

* test fixes

* win compat

* win compat again?

* size_t arg fixup

* casts

* size_t fixes

* more size_t fix

* json fixes

* builtin Py_UNUSED

Author: WillAyd

meson.build

@@ -7,7 +7,8 @@ project(
     meson_version: '>=1.2.1',
     default_options: [
         'buildtype=release',
-        'c_std=c11'
+        'c_std=c11',
+        'warning_level=2',
     ]
 )
 

pandas/_libs/include/pandas/datetime/pd_datetime.h

@@ -50,7 +50,7 @@ typedef struct {
                                  NPY_DATETIMEUNIT *, int *, int *, const char *,
                                  int, FormatRequirement);
   int (*get_datetime_iso_8601_strlen)(int, NPY_DATETIMEUNIT);
-  int (*make_iso_8601_datetime)(npy_datetimestruct *, char *, int, int,
+  int (*make_iso_8601_datetime)(npy_datetimestruct *, char *, size_t, int,
                                 NPY_DATETIMEUNIT);
   int (*make_iso_8601_timedelta)(pandas_timedeltastruct *, char *, size_t *);
 } PandasDateTime_CAPI;

pandas/_libs/include/pandas/portable.h

@@ -23,3 +23,15 @@ The full license is in the LICENSE file, distributed with this software.
 #define isspace_ascii(c) (((c) == ' ') || (((unsigned)(c) - '\t') < 5))
 #define toupper_ascii(c) ((((unsigned)(c) - 'a') < 26) ? ((c) & 0x5f) : (c))
 #define tolower_ascii(c) ((((unsigned)(c) - 'A') < 26) ? ((c) | 0x20) : (c))
+
+#if defined(_WIN32)
+#define PD_FALLTHROUGH                                                         \
+  do {                                                                         \
+  } while (0) /* fallthrough */
+#elif __has_attribute(__fallthrough__)
+#define PD_FALLTHROUGH __attribute__((__fallthrough__))
+#else
+#define PD_FALLTHROUGH                                                         \
+  do {                                                                         \
+  } while (0) /* fallthrough */
+#endif

pandas/_libs/include/pandas/vendored/numpy/datetime/np_datetime_strings.h

@@ -85,7 +85,7 @@ int get_datetime_iso_8601_strlen(int local, NPY_DATETIMEUNIT base);
  *  Returns 0 on success, -1 on failure (for example if the output
  *  string was too short).
  */
-int make_iso_8601_datetime(npy_datetimestruct *dts, char *outstr, int outlen,
+int make_iso_8601_datetime(npy_datetimestruct *dts, char *outstr, size_t outlen,
                            int utc, NPY_DATETIMEUNIT base);
 
 /*

pandas/_libs/src/datetime/pd_datetime.c

@@ -21,6 +21,7 @@ This file is derived from NumPy 1.7. See NUMPY_LICENSE.txt
 
 #include "datetime.h"
 #include "pandas/datetime/pd_datetime.h"
+#include "pandas/portable.h"
 
 static void pandas_datetime_destructor(PyObject *op) {
   void *ptr = PyCapsule_GetPointer(op, PandasDateTime_CAPSULE_NAME);
@@ -188,7 +189,7 @@ static npy_datetime PyDateTimeToEpoch(PyObject *dt, NPY_DATETIMEUNIT base) {
   return npy_dt;
 }
 
-static int pandas_datetime_exec(PyObject *module) {
+static int pandas_datetime_exec(PyObject *Py_UNUSED(module)) {
   PyDateTime_IMPORT;
   PandasDateTime_CAPI *capi = PyMem_Malloc(sizeof(PandasDateTime_CAPI));
   if (capi == NULL) {

pandas/_libs/src/parser/pd_parser.c

@@ -100,7 +100,7 @@ static void pandas_parser_destructor(PyObject *op) {
   PyMem_Free(ptr);
 }
 
-static int pandas_parser_exec(PyObject *module) {
+static int pandas_parser_exec(PyObject *Py_UNUSED(module)) {
   PandasParser_CAPI *capi = PyMem_Malloc(sizeof(PandasParser_CAPI));
   if (capi == NULL) {
     PyErr_NoMemory();

pandas/_libs/src/parser/tokenizer.c

@@ -795,7 +795,7 @@ static int tokenize_bytes(parser_t *self, size_t line_limit,
         break;
       } else if (!isblank(c)) {
         self->state = START_FIELD;
-        // fall through to subsequent state
+        PD_FALLTHROUGH; // fall through to subsequent state
       } else {
         // if whitespace char, keep slurping
         break;
@@ -849,12 +849,12 @@ static int tokenize_bytes(parser_t *self, size_t line_limit,
           self->state = WHITESPACE_LINE;
           break;
         }
-        // fall through
       }
 
       // normal character - fall through
       // to handle as START_FIELD
       self->state = START_FIELD;
+      PD_FALLTHROUGH;
     }
     case START_FIELD:
       // expecting field
@@ -1130,10 +1130,10 @@ int parser_consume_rows(parser_t *self, size_t nrows) {
 
   /* if word_deletions == 0 (i.e. this case) then char_count must
    * be 0 too, as no data needs to be skipped */
-  const int64_t char_count = word_deletions >= 1
-                                 ? (self->word_starts[word_deletions - 1] +
-                                    strlen(self->words[word_deletions - 1]) + 1)
-                                 : 0;
+  const uint64_t char_count =
+      word_deletions >= 1 ? (self->word_starts[word_deletions - 1] +
+                             strlen(self->words[word_deletions - 1]) + 1)
+                          : 0;
 
   TRACE(("parser_consume_rows: Deleting %d words, %d chars\n", word_deletions,
          char_count));
@@ -1415,9 +1415,11 @@ double xstrtod(const char *str, char **endptr, char decimal, char sci,
   int negative = 0;
   switch (*p) {
   case '-':
-    negative = 1; // Fall through to increment position.
+    negative = 1;
+    PD_FALLTHROUGH; // Fall through to increment position.
   case '+':
     p++;
+    break;
   }
 
   int exponent = 0;
@@ -1485,9 +1487,11 @@ double xstrtod(const char *str, char **endptr, char decimal, char sci,
     negative = 0;
     switch (*++p) {
     case '-':
-      negative = 1; // Fall through to increment pos.
+      negative = 1;
+      PD_FALLTHROUGH; // Fall through to increment position.
     case '+':
       p++;
+      break;
     }
 
     // Process string of digits.
@@ -1595,9 +1599,11 @@ double precise_xstrtod(const char *str, char **endptr, char decimal, char sci,
   int negative = 0;
   switch (*p) {
   case '-':
-    negative = 1; // Fall through to increment position.
+    negative = 1;
+    PD_FALLTHROUGH; // Fall through to increment position.
   case '+':
     p++;
+    break;
   }
 
   double number = 0.;
@@ -1656,9 +1662,11 @@ double precise_xstrtod(const char *str, char **endptr, char decimal, char sci,
     negative = 0;
     switch (*++p) {
     case '-':
-      negative = 1; // Fall through to increment pos.
+      negative = 1;
+      PD_FALLTHROUGH; // Fall through to increment position.
     case '+':
       p++;
+      break;
     }
 
     // Process string of digits.
@@ -1762,8 +1770,8 @@ static char *_str_copy_decimal_str_c(const char *s, char **endpos, char decimal,
   return s_copy;
 }
 
-double round_trip(const char *p, char **q, char decimal, char sci, char tsep,
-                  int skip_trailing, int *error, int *maybe_int) {
+double round_trip(const char *p, char **q, char decimal, char Py_UNUSED(sci),
+                  char tsep, int skip_trailing, int *error, int *maybe_int) {
   // 'normalize' representation to C-locale; replace decimal with '.' and
   // remove thousands separator.
   char *endptr;
@@ -1975,7 +1983,7 @@ uint64_t str_to_uint64(uint_state *state, const char *p_item, int64_t int_max,
         break;
       }
       if ((number < pre_max) ||
-          ((number == pre_max) && (d - '0' <= dig_pre_max))) {
+          ((number == pre_max) && ((uint64_t)(d - '0') <= dig_pre_max))) {
         number = number * 10 + (d - '0');
         d = *++p;
 
@@ -1987,7 +1995,7 @@ uint64_t str_to_uint64(uint_state *state, const char *p_item, int64_t int_max,
   } else {
     while (isdigit_ascii(d)) {
       if ((number < pre_max) ||
-          ((number == pre_max) && (d - '0' <= dig_pre_max))) {
+          ((number == pre_max) && ((uint64_t)(d - '0') <= dig_pre_max))) {
         number = number * 10 + (d - '0');
         d = *++p;