[modules] Introduce putStringConstantLexem()

thoni56 · Aug 22, 2023 · b6daa62 · b6daa62
1 parent ff3b7eb
commit b6daa62
Show file tree

Hide file tree

Showing 4 changed files with 50 additions and 47 deletions.
diff --git a/src/lexembuffer.c b/src/lexembuffer.c
@@ -5,6 +5,7 @@
 #include "commons.h"
 #include "globals.h"
 #include "lexem.h"
+#include "options.h"
 
 //
 // Basic manipulation, these should preferably be private in favour
@@ -200,6 +201,51 @@ int putIdentifierLexem(LexemBuffer *lexemBuffer, CharacterBuffer *characterBuffe
     return ch;
 }
 
+void putStringConstantLexem(LexemBuffer *lb, CharacterBuffer *cb, int lexemStartingColumn) {
+    int ch;
+    int line = lineNumberFrom(cb);
+    int size = 0;
+
+    putLexemCode(lb, STRING_LITERAL);
+    do {
+        ch = getChar(cb);
+        size++;
+        if (ch != '\"'
+            && size < MAX_LEXEM_SIZE - 10) /* WTF is 10? Perhaps required space for position info? */
+            putLexemChar(lb, ch);
+        if (ch == '\\') {
+            ch = getChar(cb);
+            size++;
+            if (size < MAX_LEXEM_SIZE - 10)
+                putLexemChar(lb, ch);
+            /* TODO escape sequences */
+            if (ch == '\n') {
+                cb->lineNumber++;
+                cb->lineBegin    = cb->nextUnread;
+                cb->columnOffset = 0;
+            }
+            continue;
+        }
+        if (ch == '\n') {
+            cb->lineNumber++;
+            cb->lineBegin    = cb->nextUnread;
+            cb->columnOffset = 0;
+            if (options.strictAnsi && (options.debug || options.errors)) {
+                warningMessage(ERR_ST, "string constant through end of line");
+            }
+        }
+        // in Java CR LF can't be a part of string, even there
+        // are benchmarks making Xrefactory coredump if CR or LF
+        // is a part of strings
+    } while (ch != '\"' && (ch != '\n' || !options.strictAnsi) && ch != -1);
+    if (ch == -1 && options.mode != ServerMode) {
+        warningMessage(ERR_ST, "string constant through EOF");
+    }
+    terminateLexemString(lb);
+    putLexemPositionFields(lb, fileNumberFrom(cb), lineNumberFrom(cb), lexemStartingColumn);
+    putLexemLines(lb, lineNumberFrom(cb) - line);
+}
+
 void putCharLiteralLexem(LexemBuffer *lb, CharacterBuffer *cb, int lexemStartingColumn,
                          int length, unsigned chval) {
     putLexemCode(lb, CHAR_LITERAL);

diff --git a/src/lexembuffer.h b/src/lexembuffer.h
@@ -38,7 +38,6 @@ extern void moveLexemStreamWriteToBackpatchPositonWithOffset(LexemBuffer *lb, in
 extern int strlenOfBackpatchedIdentifier(LexemBuffer *lb);
 
 /* Put elementary values */
-extern void putLexemChar(LexemBuffer *lb, char ch);
 extern void putLexemCode(LexemBuffer *lb, LexemCode lexem);
 
 /* Put semantically meaningful complete lexems including position, string, value, ... */
@@ -53,6 +52,7 @@ extern void putIntegerLexem(LexemBuffer *lb, LexemCode lexem, long unsigned valu
 extern void putCompletionLexem(LexemBuffer *lb, CharacterBuffer *cb, int len);
 extern void putFloatingPointLexem(LexemBuffer *lb, LexemCode lexem, CharacterBuffer *cb,
                                   int lexemStartingColumn, int lexStartFilePos);
+extern void putStringConstantLexem(LexemBuffer *lb, CharacterBuffer *cb, int lexemStartingColumn);
 extern void putCharLiteralLexem(LexemBuffer *lb, CharacterBuffer *cb, int lexemStartingColumn,
                                 int length, unsigned chval);
 extern void terminateLexemString(LexemBuffer *lb);

diff --git a/src/lexembuffer_tests.c b/src/lexembuffer_tests.c
@@ -9,6 +9,7 @@
 #include "commons.mock"
 #include "fileio.mock"
 #include "zlib.mock"
+#include "options.mock"
 
 
 static LexemBuffer lb;
@@ -175,6 +176,7 @@ Ensure(LexemBuffer, can_put_and_get_position) {
     assert_that(lb.read, is_equal_to(pointer_after_put));
 }
 
+extern void putLexemChar(LexemBuffer *lb, char ch);
 Ensure(LexemBuffer, can_backpatch_lexem) {
     LexemCode backpatched_lexem = STRING_LITERAL;
 

diff --git a/src/lexer.c b/src/lexer.c
@@ -290,51 +290,6 @@ static int scanIntegerValue(CharacterBuffer *cb, int ch, unsigned long *valueP)
     return ch;
 }
 
-static void handleStringConstant(CharacterBuffer *cb, LexemBuffer *lb, int lexemStartingColumn) {
-    int ch;
-    int line = lineNumberFrom(cb);
-    int size = 0;
-
-    putLexemCode(lb, STRING_LITERAL);
-    do {
-        ch = getChar(cb);
-        size++;
-        if (ch != '\"'
-            && size < MAX_LEXEM_SIZE - 10) /* WTF is 10? Perhaps required space for position info? */
-            putLexemChar(lb, ch);
-        if (ch == '\\') {
-            ch = getChar(cb);
-            size++;
-            if (size < MAX_LEXEM_SIZE - 10)
-                putLexemChar(lb, ch);
-            /* TODO escape sequences */
-            if (ch == '\n') {
-                cb->lineNumber++;
-                cb->lineBegin    = cb->nextUnread;
-                cb->columnOffset = 0;
-            }
-            continue;
-        }
-        if (ch == '\n') {
-            cb->lineNumber++;
-            cb->lineBegin    = cb->nextUnread;
-            cb->columnOffset = 0;
-            if (options.strictAnsi && (options.debug || options.errors)) {
-                warningMessage(ERR_ST, "string constant through end of line");
-            }
-        }
-        // in Java CR LF can't be a part of string, even there
-        // are benchmarks making Xrefactory coredump if CR or LF
-        // is a part of strings
-    } while (ch != '\"' && (ch != '\n' || !options.strictAnsi) && ch != -1);
-    if (ch == -1 && options.mode != ServerMode) {
-        warningMessage(ERR_ST, "string constant through EOF");
-    }
-    putLexemChar(lb, 0); /* Terminate string */
-    putLexemPositionFields(lb, fileNumberFrom(cb), lineNumberFrom(cb), lexemStartingColumn);
-    putLexemLines(lb, lineNumberFrom(cb) - line);
-}
-
 static int handleCharConstant(CharacterBuffer *cb, LexemBuffer *lb, int lexemStartingColumn) {
     int      fileOffsetForLexemStart, ch;
     unsigned chval = 0;
@@ -732,7 +687,7 @@ bool buildLexemFromCharacters(CharacterBuffer *cb, LexemBuffer *lb) {
             }
 
             case '\"':
-                handleStringConstant(cb, lb, lexemStartingColumn);
+                putStringConstantLexem(lb, cb, lexemStartingColumn);
                 ch = getChar(cb);
                 goto nextLexem;