impr: regexp - print the actual error message if compilation fails (#104

)
nalgeon · Feb 2, 2024 · 84671a0 · 84671a0
1 parent 7a59981
commit 84671a0
Show file tree

Hide file tree

Showing 3 changed files with 45 additions and 5 deletions.
diff --git a/src/regexp/extension.c b/src/regexp/extension.c
@@ -62,7 +62,9 @@ static void regexp_statement(sqlite3_context* context, int argc, sqlite3_value**
     if (re == NULL) {
         re = regexp.compile(pattern);
         if (re == NULL) {
-            sqlite3_result_error_nomem(context);
+            char* msg = regexp.get_error(pattern);
+            sqlite3_result_error(context, msg, -1);
+            free(msg);
             return;
         }
         is_new_re = true;
@@ -121,7 +123,9 @@ static void regexp_like(sqlite3_context* context, int argc, sqlite3_value** argv
     if (re == NULL) {
         re = regexp.compile(pattern);
         if (re == NULL) {
-            sqlite3_result_error_nomem(context);
+            char* msg = regexp.get_error(pattern);
+            sqlite3_result_error(context, msg, -1);
+            free(msg);
             return;
         }
         is_new_re = true;
@@ -171,7 +175,9 @@ static void regexp_substr(sqlite3_context* context, int argc, sqlite3_value** ar
     if (re == NULL) {
         re = regexp.compile(pattern);
         if (re == NULL) {
-            sqlite3_result_error_nomem(context);
+            char* msg = regexp.get_error(pattern);
+            sqlite3_result_error(context, msg, -1);
+            free(msg);
             return;
         }
         is_new_re = true;
@@ -243,7 +249,9 @@ static void regexp_capture(sqlite3_context* context, int argc, sqlite3_value** a
     if (re == NULL) {
         re = regexp.compile(pattern);
         if (re == NULL) {
-            sqlite3_result_error_nomem(context);
+            char* msg = regexp.get_error(pattern);
+            sqlite3_result_error(context, msg, -1);
+            free(msg);
             return;
         }
         is_new_re = true;
@@ -309,7 +317,9 @@ static void regexp_replace(sqlite3_context* context, int argc, sqlite3_value** a
     if (re == NULL) {
         re = regexp.compile(pattern);
         if (re == NULL) {
-            sqlite3_result_error_nomem(context);
+            char* msg = regexp.get_error(pattern);
+            sqlite3_result_error(context, msg, -1);
+            free(msg);
             return;
         }
         is_new_re = true;

diff --git a/src/regexp/regexp.c b/src/regexp/regexp.c
@@ -32,6 +32,34 @@ static void re_free(pcre2_code* re) {
     pcre2_code_free(re);
 }
 
+/*
+ * re_get_error returns the error message for a given pattern.
+ */
+static char* re_get_error(const char* pattern) {
+    size_t erroffset;
+    int errcode;
+    uint32_t options = PCRE2_UCP | PCRE2_UTF;
+    pcre2_code* re = pcre2_compile((PCRE2_SPTR8)pattern, PCRE2_ZERO_TERMINATED, options, &errcode,
+                                   &erroffset, NULL);
+
+    if (re != NULL) {
+        // free the compiled pattern if successful
+        pcre2_code_free(re);
+        return NULL;
+    }
+
+    PCRE2_UCHAR buffer[256];
+    pcre2_get_error_message(errcode, buffer, sizeof(buffer));
+
+    // Allocate memory for the error message
+    // (additional space for formatting)
+    char* msg = (char*)malloc(256 + 32);
+    if (msg != NULL) {
+        snprintf(msg, 256 + 32, "%s (offset %d)", buffer, (int)erroffset);
+    }
+    return msg;
+}
+
 /*
  * re_like checks if source string matches pattern.
  * returns:
@@ -143,6 +171,7 @@ static int re_replace(pcre2_code* re, const char* source, const char* repl, char
 struct regexp_ns regexp = {
     .compile = re_compile,
     .free = re_free,
+    .get_error = re_get_error,
     .like = re_like,
     .extract = re_extract,
     .replace = re_replace,

diff --git a/src/regexp/regexp.h b/src/regexp/regexp.h
@@ -9,6 +9,7 @@
 struct regexp_ns {
     pcre2_code* (*compile)(const char* pattern);
     void (*free)(pcre2_code* re);
+    char* (*get_error)(const char* pattern);
     int (*like)(pcre2_code* re, const char* source);
     int (*extract)(pcre2_code* re, const char* source, size_t group_idx, char** substr);
     int (*replace)(pcre2_code* re, const char* source, const char* repl, char** dest);