[osh] Implement shopt -s nocasematch (#1748)

All tests in spec/nocasematch-match/test.sh now pass (3 failures previously allowed).

[translation] Use #included macros directly.  We should move away from this style with mycpp/yaks "modules"

---------

Co-authored-by: Andy C <[email protected]>

cpp/libc.cc

@@ -36,16 +36,16 @@ BigStr* realpath(BigStr* path) {
   return result;
 }
 
-int fnmatch(BigStr* pat, BigStr* str) {
+int fnmatch(BigStr* pat, BigStr* str, int flags) {
   // TODO: We should detect this at ./configure time, and then maybe flag these
   // at parse time, not runtime
 #ifdef FNM_EXTMATCH
-  int flags = FNM_EXTMATCH;
+  int flags_todo = FNM_EXTMATCH;
 #else
-  int flags = 0;
+  int flags_todo = 0;
 #endif
 
-  int result = ::fnmatch(pat->data_, str->data_, flags);
+  int result = ::fnmatch(pat->data_, str->data_, flags_todo);
   switch (result) {
   case 0:
     return 1;
@@ -106,7 +106,7 @@ List<BigStr*>* glob(BigStr* pat) {
 
 // Raises RuntimeError if the pattern is invalid.  TODO: Use a different
 // exception?
-List<BigStr*>* regex_match(BigStr* pattern, BigStr* str) {
+List<BigStr*>* regex_match(BigStr* pattern, BigStr* str, int flags) {
   List<BigStr*>* results = NewList<BigStr*>();
 
   regex_t pat;

cpp/libc.h

@@ -20,14 +20,14 @@ BigStr* realpath(BigStr* path);
 
 BigStr* gethostname();
 
-int fnmatch(BigStr* pat, BigStr* str);
+int fnmatch(BigStr* pat, BigStr* str, int flags = 0);
 
 List<BigStr*>* glob(BigStr* pat);
 
 Tuple2<int, int>* regex_first_group_match(BigStr* pattern, BigStr* str,
                                           int pos);
 
-List<BigStr*>* regex_match(BigStr* pattern, BigStr* str);
+List<BigStr*>* regex_match(BigStr* pattern, BigStr* str, int flags = 0);
 
 int wcswidth(BigStr* str);
 int get_terminal_width();

cpp/preamble.h

@@ -2,6 +2,8 @@
 
 #include <errno.h>
 #include <fcntl.h>     // e.g. F_DUPFD used directly
+#include <fnmatch.h>   // FNM_CASEFOLD in osh/sh_expr_eval.py
+#include <regex.h>     // REG_ICASE in osh/sh_expr_eval.py
 #include <sys/wait.h>  // e.g. WIFSIGNALED() called directly
 
 #include "_gen/core/optview.h"

frontend/option_def.py

@@ -232,7 +232,6 @@ def DoneWithImplementedOptions(self):
     'mailwarn',
     'no_empty_cmd_completion',
     'nocaseglob',
-    'nocasematch',
     'progcomp_alias',
     'promptvars',
     'restricted_shell',
@@ -278,6 +277,7 @@ def _Init(opt_def):
     # shopt options that aren't in any groups.
     opt_def.Add('failglob')
     opt_def.Add('extglob')
+    opt_def.Add('nocasematch')
 
     # Compatibility
     opt_def.Add(

osh/cmd_eval.py

@@ -89,6 +89,8 @@
 
 import posix_ as posix
 import libc  # for fnmatch
+# Import this name directly because the C++ translation uses macros literally.
+from libc import FNM_CASEFOLD
 
 from typing import List, Dict, Tuple, Optional, Any, cast, TYPE_CHECKING
 
@@ -1436,6 +1438,7 @@ def _DoCase(self, node):
         # type: (command.Case) -> int
 
         to_match = self._EvalCaseArg(node.to_match, node.case_kw)
+        fnmatch_flags = FNM_CASEFOLD if self.exec_opts.nocasematch() else 0
         self._MaybeRunDebugTrap()
 
         status = 0  # If there are no arms, it should be zero?
@@ -1454,7 +1457,7 @@ def _DoCase(self, node):
                         word_val = self.word_ev.EvalWordToString(
                             pat_word, word_eval.QUOTE_FNMATCH)
 
-                        if libc.fnmatch(word_val.s, to_match_str.s):
+                        if libc.fnmatch(word_val.s, to_match_str.s, fnmatch_flags):
                             status = self._ExecuteList(case_arm.action)
                             matched = True  # TODO: Parse ;;& and for fallthrough and such?
                             break

osh/sh_expr_eval.py

@@ -57,6 +57,8 @@
 from osh import word_eval
 
 import libc  # for fnmatch
+# Import these names directly because the C++ translation uses macros literally.
+from libc import FNM_CASEFOLD, REG_ICASE
 
 from typing import Tuple, Optional, cast, TYPE_CHECKING
 if TYPE_CHECKING:
@@ -1044,14 +1046,15 @@ def EvalB(self, node):
                     raise AssertionError(op_id)  # should never happen
 
                 if arg_type == bool_arg_type_e.Str:
+                    fnmatch_flags = FNM_CASEFOLD if self.exec_opts.nocasematch() else 0
 
                     if op_id in (Id.BoolBinary_GlobEqual,
                                  Id.BoolBinary_GlobDEqual):
                         #log('Matching %s against pattern %s', s1, s2)
-                        return libc.fnmatch(s2, s1)
+                        return libc.fnmatch(s2, s1, fnmatch_flags)
 
                     if op_id == Id.BoolBinary_GlobNEqual:
-                        return not libc.fnmatch(s2, s1)
+                        return not libc.fnmatch(s2, s1, fnmatch_flags)
 
                     if op_id in (Id.BoolBinary_Equal, Id.BoolBinary_DEqual):
                         return s1 == s2
@@ -1062,8 +1065,10 @@ def EvalB(self, node):
                     if op_id == Id.BoolBinary_EqualTilde:
                         # TODO: This should go to --debug-file
                         #log('Matching %r against regex %r', s1, s2)
+                        regex_flags = REG_ICASE if self.exec_opts.nocasematch() else 0
+
                         try:
-                            matches = libc.regex_match(s2, s1)
+                            matches = libc.regex_match(s2, s1, regex_flags)
                         except RuntimeError as e:
                             # Status 2 indicates a regex parse error.  This is fatal in OSH but
                             # not in bash, which treats [[ like a command with an exit code.

pyext/libc.c

@@ -55,12 +55,12 @@ static PyObject *
 func_fnmatch(PyObject *self, PyObject *args) {
   const char *pattern;
   const char *str;
+  int flags = 0;
 
-  if (!PyArg_ParseTuple(args, "ss", &pattern, &str)) {
+  if (!PyArg_ParseTuple(args, "ss|i", &pattern, &str, &flags)) {
     return NULL;
   }
 
-  int flags = 0;
   // NOTE: Testing for __GLIBC__ is the version detection anti-pattern.  We
   // should really use feature detection in our configure script.  But I plan
   // to get rid of the dependency on FNM_EXTMATCH because it doesn't work on
@@ -185,12 +185,15 @@ static PyObject *
 func_regex_match(PyObject *self, PyObject *args) {
   const char* pattern;
   const char* str;
-  if (!PyArg_ParseTuple(args, "ss", &pattern, &str)) {
+  int flags = 0;
+
+  if (!PyArg_ParseTuple(args, "ss|i", &pattern, &str, &flags)) {
     return NULL;
   }
 
+  flags |= REG_EXTENDED;
   regex_t pat;
-  int status = regcomp(&pat, pattern, REG_EXTENDED);
+  int status = regcomp(&pat, pattern, flags);
   if (status != 0) {
     char error_string[80];
     regerror(status, &pat, error_string, 80);
@@ -408,7 +411,14 @@ static PyMethodDef methods[] = {
 #endif
 
 void initlibc(void) {
-  Py_InitModule("libc", methods);
+  PyObject *module;
+
+  module = Py_InitModule("libc", methods);
+  if (module != NULL) {
+      PyModule_AddIntConstant(module, "FNM_CASEFOLD", FNM_CASEFOLD);
+      PyModule_AddIntConstant(module, "REG_ICASE", REG_ICASE);
+  }
+
   errno_error = PyErr_NewException("libc.error",
                                     PyExc_IOError, NULL);
 }

pyext/libc.pyi

@@ -1,10 +1,13 @@
 from typing import List, Optional, Tuple
 
+FNM_CASEFOLD: int
+REG_ICASE: int
+
 def gethostname() -> str: ...
 def glob(pat: str) -> List[str]: ...
-def fnmatch(pat: str, s: str) -> bool: ...
+def fnmatch(pat: str, s: str, flags: int = 0) -> bool: ...
 def regex_first_group_match(regex: str, s: str, pos: int) -> Optional[Tuple[int, int]]: ...
-def regex_match(regex: str, s: str) -> Optional[List[str]]: ...
+def regex_match(regex: str, s: str, flags: int = 0) -> Optional[List[str]]: ...
 def wcswidth(s: str) -> int: ...
 def get_terminal_width() -> int: ...
 def print_time(real: float, user: float, sys: float) -> None: ...

spec/nocasematch-match.test.sh

@@ -1,5 +1,5 @@
 ## compare_shells: bash
-## oils_failures_allowed: 3
+## oils_failures_allowed: 0
 
 # Tests nocasematch matching