htex

commit 0544bcf9a77afced6fe52c018fb1c4758ffde18c
parent f527a413ca2f869e114bee4d0b3db7f55dbfaf21
Author: Robin <kroekerrobin@gmail.com>
Date:   Sun, 13 Aug 2023 16:23:41 +0200

Awake innerText to life

Diffstat:
M
htex.c
 | 
14
+++++++++++++-
M
html.c
 | 
56
++++++++++++++++++++++++++++++++++++++++++++++++++++----
M
html.h
 | 
4
+++-
M
lib.c
 | 
3
+--

4 files changed, 69 insertions(+), 8 deletions(-)
diff --git a/htex.c b/htex.c
@@ -133,24 +133,34 @@ int main(int argc, char *argv[])
 	int o = 0;
 	int option_index = 0;
 	bool isInnerHtml = false;
+	bool isInnerText = false;
 	bool isExcept = false;
 	char *text = NULL;
 	char *searchPattern = NULL;
 	static struct option long_options[] = {
 		{ "innerhtml", no_argument, 0, 'i' },
+		{ "innertext", no_argument, 0, 't' },
 		{ "except", no_argument, 0, 'e' },
 		{ 0, 0, 0, 0 }
 	};
-	while ((o = getopt_long(argc, argv, "ie", long_options, &option_index)) != -1) {
+	while ((o = getopt_long(argc, argv, "ite", long_options, &option_index)) != -1) {
 		switch(o) {
 			case 'i':
 				isInnerHtml = true;
 				break;
+			case 't':
+				isInnerText = true;
+				break;
 			case 'e':
 				isExcept = true;
 				break;
 		}
 	}
+	if (isInnerHtml && isInnerText)
+	{
+		fprintf(stderr, "Provide either --innerhtml or --innertext.\n");
+		return -1;
+	}
 	if (argc == optind)
 	{
 		fprintf(stderr, "Provide a search pattern!\n");
@@ -188,6 +198,8 @@ int main(int argc, char *argv[])
 	options->isExcept = isExcept;
 	if (isInnerHtml)
 		options->out = OUT_INNER_HTML;
+	if (isInnerText)
+		options->out = OUT_INNER_TEXT;
 	filterHtml(text, options);
 	freeOpts(options);
 	free(text);
diff --git a/html.c b/html.c
@@ -36,6 +36,8 @@ struct tag *initTag()
 	struct tag *t = malloc(sizeof(struct tag));
 	t->name = malloc(sizeof(char));
 	t->name[0] = 0;
+	t->innerText = malloc(sizeof(char));
+	t->innerText[0] = 0;
 	t->attrs = NULL;
 	t->children = NULL;
 	t->attrsLen = 0;
@@ -325,6 +327,8 @@ struct tag *parseTag(char *text, size_t offset, enum state state, struct tag_lis
 						state = STATE_TAG;
 						break;
 					}
+					stillOpenTag = getLastOpenTag(tagList);
+					stillOpenTag->innerText = stringCat(stillOpenTag->innerText, cpToChars(cp, ret));
 					break;
 				case STATE_TAG:
 					if (cp == SOLIDUS)
@@ -545,6 +549,7 @@ struct tag *parseTag(char *text, size_t offset, enum state state, struct tag_lis
 void freeTag(struct tag *t)
 {
 	free(t->name);
+	free(t->innerText);
 	for (int i=0; i<t->attrsLen; i++)
 	{
 		free(t->attrs[i]->name);
@@ -655,7 +660,32 @@ void printHtml(struct tag *t, int indent)
 
 void printTag(char *text, struct tag *t, enum output_type out, struct tag_list *foundTags)
 {
-	unsigned int p = *(unsigned int *)t;
+	// bool doPrint = true;
+	switch (out)
+	{
+		case OUT_INNER_HTML:
+			/* for (int i=t->_innerHtmlBeginOffset; i<t->_innerHtmlEndOffset; i++)
+			{
+				doPrint = true;
+				for (int k=0; k<foundTags->len; k++)
+				{
+					if (
+						foundTags->tags[k]->_innerHtmlBeginOffset < i &&
+						foundTags->tags[k]->_innerHtmlEndOffset > i
+					)
+						doPrint = false;
+				}
+				if (doPrint)
+					putchar(text[i]);
+			} */
+			break;
+		case OUT_OUTER_HTML:
+			break;
+		case OUT_INNER_TEXT:
+			break;
+	}
+	putchar('\n');
+	/* unsigned int p = *(unsigned int *)t;
 	unsigned int cp;
 	bool isMatch = false;
 	for (int i=0; i<foundTags->len; i++)
@@ -679,7 +709,7 @@ void printTag(char *text, struct tag *t, enum output_type out, struct tag_list *
 		if (strlen(trimmedText) > 0)
 			printf("%s\n", trimmedText);
 		free(trimmedText);
-	}
+	} */
 	for (int i=t->childrenLen-1; i>-1; i--)
 	{
 		printTag(text, t->children[i], out, foundTags);
@@ -695,19 +725,37 @@ void printResult
 )
 {
 	if (opts->isExcept)
+	{
+		/* printf("len: %ld\n", foundTags->len);
+		char *trimmedText = NULL;
+		for (int i=0; i<foundTags->len; i++)
+		{
+			trimmedText = trim(getOuterHtml(text, foundTags->tags[i]));
+			printf("%s\n", trimmedText);
+			free(trimmedText);
+		} */
 		printTag(text, rootTag, opts->out, foundTags);
+	}
 	else
 	{
+		char *requestedText = NULL;
 		char *trimmedText = NULL;
 		for (int i=0; i<foundTags->len; i++)
 		{
 			switch (opts->out)
 			{
 				case OUT_INNER_HTML:
-					trimmedText = trim(getInnerHtml(text, foundTags->tags[i]));
+					requestedText = getInnerHtml(text, foundTags->tags[i]);
+					trimmedText = trim(requestedText);
+					free(requestedText);
 					break;
 				case OUT_OUTER_HTML:
-					trimmedText = trim(getOuterHtml(text, foundTags->tags[i]));
+					requestedText = getOuterHtml(text, foundTags->tags[i]);
+					trimmedText = trim(requestedText);
+					free(requestedText);
+					break;
+				case OUT_INNER_TEXT:
+					trimmedText = trim(foundTags->tags[i]->innerText);
 					break;
 			}
 			if (strlen(trimmedText) > 0)
diff --git a/html.h b/html.h
@@ -23,7 +23,8 @@ const char *voidElements[] = {
 enum output_type
 {
 	OUT_INNER_HTML,
-	OUT_OUTER_HTML
+	OUT_OUTER_HTML,
+	OUT_INNER_TEXT
 };
 
 struct filter_opts
@@ -46,6 +47,7 @@ struct tag
 	char *name;
 	struct attr **attrs;
 	struct tag **children;
+	char *innerText;
 	size_t attrsLen;
 	size_t childrenLen;
 	bool _isVoidElement; // means there is no closing tag
diff --git a/lib.c b/lib.c
@@ -1,4 +1,4 @@
-char *stringCat(char *str1,char *str2)
+char *stringCat(char *str1, char *str2)
 {
 	int str1Len = 0;
 	int str2Len = 0;
@@ -74,7 +74,6 @@ char *trim(char *text)
 	}
 	trimmedText = realloc(trimmedText, (k+1) * sizeof(char));
 	trimmedText[k] = 0;
-	free(text);
 	return trimmedText;
 }