htex

commit 27e1b9ab0d7cab9372e897a59f6c81aa3b7b2e6b
parent d847e830279219425e57ef45bde04eed13f2ad4b
Author: Robin <kroekerrobin@gmail.com>
Date:   Tue,  8 Aug 2023 22:33:15 +0200

Trim text before outputting

Diffstat:
M
htex.c
 | 
8
++------
M
html.c
 | 
18
+++++++++++++-----
M
html.h
 | 
16
+++++++---------
M
lib.c
 | 
46
++++++++++++++++++++++++++++++++++++++++++++++

4 files changed, 68 insertions(+), 20 deletions(-)
diff --git a/htex.c b/htex.c
@@ -45,8 +45,7 @@ char *readFile(FILE *fp)
 struct filter_opts *parseFilterOpts(const char *pattern)
 {
 	struct filter_opts *opt = malloc(sizeof(struct filter_opts));
-	opt->outerHtml = true;
-	opt->innerHtml = false;
+	opt->out = OUT_OUTER_HTML;
 	opt->tag = malloc(sizeof(char));
 	opt->tag[0] = 0;
 	opt->attr = malloc(sizeof(char));
@@ -224,10 +223,7 @@ int main(int argc, char *argv[])
 	}
 	struct filter_opts *options = parseFilterOpts(searchPattern);
 	if (isInnerHtml)
-	{
-		options->innerHtml = true;
-		options->outerHtml = false;
-	}
+		options->out = OUT_INNER_HTML;
 	filterHtml(text, options);
 	freeOpts(options);
 	free(text);
diff --git a/html.c b/html.c
@@ -566,7 +566,7 @@ struct tag *findTag(struct tag *tag, struct tag_list *list, struct filter_opts *
 		if (matchesAttrValue)
 			return tag;
 	}
-	for (int i=0; i<tag->childrenLen; i++)
+	for (int i=tag->childrenLen-1; i>-1; i--)
 	{
 		struct tag *foundTag = findTag(tag->children[i], list, opt);
 		if (foundTag != NULL)
@@ -591,12 +591,20 @@ void filterHtml(char *text, struct filter_opts *opts)
 		printError("No tag found.");
 	else
 	{
-		printf("result: %s\n", result->name);
-		if (!result->_isVoidElement)
+		char *trimmedOutput;
+		if (result->_isVoidElement)
+			opts->out = OUT_OUTER_HTML;
+		switch (opts->out)
 		{
-			printf("innerHtml: %s\n", result->innerHtml);
+			case OUT_INNER_HTML:
+				trimmedOutput = trim(result->innerHtml);
+				break;
+			case OUT_OUTER_HTML:
+				trimmedOutput = trim(result->outerHtml);
+				break;
 		}
-		printf("outerHtml: %s\n", result->outerHtml);
+		printf("%s\n", trimmedOutput);
+		free(trimmedOutput);
 	}
 	freeTag(rootTag);
 	freeTagList(tagList);
diff --git a/html.h b/html.h
@@ -20,13 +20,18 @@ const char *voidElements[] = {
 	"input", "link", "meta", "source", "track", "wbr"
 };
 
+enum output_type
+{
+	OUT_INNER_HTML,
+	OUT_OUTER_HTML
+};
+
 struct filter_opts
 {
 	char *tag;
 	char *attr;
 	char *key;
-	bool innerHtml;
-	bool outerHtml;
+	enum output_type out;
 };
 
 struct attr
@@ -77,10 +82,3 @@ enum attr_value_syntax
 	AVS_APOSTROPHE,
 	AVS_UNQUOTED
 };
-
-/* enum search_type
-{
-	ST_NO,
-	ST_LIST,
-	ST_HIERARCHY
-}; */
diff --git a/lib.c b/lib.c
@@ -30,3 +30,49 @@ char *cpToChars(uint_least32_t cp, size_t len)
 	str[len] = 0;
 	return str;
 }
+
+char *trim(char *text)
+{
+	char *trimmedText = NULL;
+	int begin = 0;
+	int end = 0;
+	for (int i=0; i<strlen(text); i++)
+	{
+		if
+		(
+				text[i] == ' ' ||
+				text[i] == '\n' ||
+				text[i] == '\t' ||
+				text[i] == '\r'
+		)
+			begin++;
+		else
+			break;
+	}
+	for (int i=strlen(text)-1; i>=0; i--)
+	{
+		if
+		(
+			text[i] == ' '||
+			text[i] == '\n' ||
+			text[i] == '\t' ||
+			text[i] == '\r'
+		)
+			end++;
+		else
+			break;
+	}
+	int k = 0;
+	for (int i=0; i<strlen(text); i++)
+	{
+		if (i >= begin && i < strlen(text) - end)
+		{
+			trimmedText = realloc(trimmedText, (k+1) * sizeof(char));
+			trimmedText[k] = text[i];
+			k++;
+		}
+	}
+	trimmedText = realloc(trimmedText, (k+1) * sizeof(char));
+	trimmedText[k] = 0;
+	return trimmedText;
+}