htex

commit bcc4ec1b9dcc6d5133ad5be90a95c8de364c88df
parent 633e945a287ee1fc0b03d7bab858bbc8e7cdba09
Author: Robin <kroekerrobin@gmail.com>
Date:   Sun,  6 Aug 2023 16:28:57 +0200

Add basic tag finding

syntax: tag[key=value]

Diffstat:
M
htex.c
 | 
163
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++-------------
M
html.c
 | 
62
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++-----
M
html.h
 | 
14
++++++++++++++

3 files changed, 208 insertions(+), 31 deletions(-)
diff --git a/htex.c b/htex.c
@@ -42,38 +42,143 @@ char *readFile(FILE *fp)
 	return text;
 }
 
+struct filter_opts *parseFilterOpts(const char *pattern)
+{
+	struct filter_opts *opt = malloc(sizeof(struct filter_opts));
+	opt->tag = malloc(sizeof(char));
+	opt->tag[0] = 0;
+	opt->attr = malloc(sizeof(char));
+	opt->attr[0] = 0;
+	opt->key = malloc(sizeof(char));
+	opt->key[0] = 0;
+	char *classValue = NULL;
+	bool isClassValue = false;
+	char *idValue = NULL;
+	bool isIdValue = false;
+	int i = 0;
+	bool isAttrKey = false;
+	bool isAttrOrTag = true;
+	char *attrOrTag = NULL;
+	int aot = 0;
+	int ak = 0;
+	int av = 0;
+	switch (pattern[0])
+	{
+		case '.':
+			isClassValue = true;
+			i = 1;
+			break;
+		case '#':
+			isIdValue = true;
+			i = 1;
+			break;
+		default:
+	}
+	for (; i<strlen(pattern); i++)
+	{
+		if (pattern[i] == ']')
+			break;
+		if (
+			!isAttrKey &&
+			!isAttrOrTag &&
+			pattern[i] != ']' &&
+			pattern[i] != '"'
+		)
+		{
+			opt->attr = realloc(opt->attr, (av+1) * sizeof(char));
+			opt->attr[av] = pattern[i];
+			av++;
+		}
+		if (pattern[i] == '=')
+			isAttrKey = false;
+		if (isAttrKey && !isAttrOrTag)
+		{
+			opt->key = realloc(opt->key, (ak+1) * sizeof(char));
+			opt->key[ak] = pattern[i];
+			ak++;
+		}
+		if (pattern[i] == '[')
+		{
+			isAttrKey = true;
+			isAttrOrTag = false;
+		}
+		if (isAttrOrTag)
+		{
+			attrOrTag = realloc(attrOrTag, (aot+1) * sizeof(char));
+			attrOrTag[aot] = pattern[i];
+			aot++;
+		}
+	}
+	attrOrTag = realloc(attrOrTag, (aot+1) * sizeof(char));
+	attrOrTag[aot] = 0;
+	if (isIdValue)
+	{
+		free(opt->key);
+		opt->key = NULL;
+		free(opt->attr);
+		opt->attr = NULL;
+		opt->attr = attrOrTag;
+		opt->key = realloc(opt->key, 3 * sizeof(char));
+		opt->key[0] = 'i';
+		opt->key[1] = 'd';
+		opt->key[2] = 0;
+	}
+	else if (isClassValue)
+	{
+		free(opt->key);
+		opt->key = NULL;
+		free(opt->attr);
+		opt->attr = NULL;
+		opt->attr = attrOrTag;
+		opt->key = realloc(opt->key, 6 * sizeof(char));
+		opt->key[0] = 'c';
+		opt->key[1] = 'l';
+		opt->key[2] = 'a';
+		opt->key[3] = 's';
+		opt->key[4] = 's';
+		opt->key[5] = 0;
+	}
+	else
+	{
+		free(opt->tag);
+		opt->tag = attrOrTag;
+		if (av > 0)
+		{
+			opt->attr = realloc(opt->attr, (av+1) * sizeof(char));
+			opt->attr[av] = 0;
+		}
+		if (ak > 0)
+		{
+			opt->key = realloc(opt->key, (ak+1) * sizeof(char));
+			opt->key[ak] = 0;
+		}
+	}
+	return opt;
+}
+
+void freeOpts(struct filter_opts *opt)
+{
+	free(opt->tag);
+	free(opt->attr);
+	free(opt->key);
+	free(opt);
+}
+
 int main(int argc, char *argv[])
 {
 	int o = 0;
 	int option_index = 0;
-	char *tag = NULL;
-	char *attribute = NULL;
-	char *key = NULL;
 	bool isInnerHtml = false;
 	bool isExcept = false;
 	char *text = NULL;
+	char *searchPattern = NULL;
 	static struct option long_options[] = {
-		{ "tag", required_argument, 0, 't' },
-		{ "attribute", required_argument, 0, 'a' },
-		{ "key", required_argument, 0, 'k' },
 		{ "innerhtml", no_argument, 0, 'i' },
 		{ "except", no_argument, 0, 'e' },
 		{ 0, 0, 0, 0 }
 	};
-	while ((o = getopt_long(argc, argv, "t:a:k:ie", long_options, &option_index)) != -1) {
+	while ((o = getopt_long(argc, argv, "ie", long_options, &option_index)) != -1) {
 		switch(o) {
-			case 't':
-				tag = malloc((strlen(optarg)+1) * sizeof(char));
-				strcpy(tag, optarg);
-				break;
-			case 'a':
-				attribute = malloc((strlen(optarg)+1) * sizeof(char));
-				strcpy(attribute, optarg);
-				break;
-			case 'k':
-				key = malloc((strlen(optarg)+1) * sizeof(char));
-				strcpy(key, optarg);
-				break;
 			case 'i':
 				isInnerHtml = true;
 				break;
@@ -82,17 +187,24 @@ int main(int argc, char *argv[])
 				break;
 		}
 	}
-	if (argc > optind+1)
+	if (argc == optind)
+	{
+		fprintf(stderr, "Provide a search pattern!\n");
+		return -1;
+	}
+	if (argc > optind+2)
 	{
 		fprintf(stderr, "Provide only one file!\n");
 		return -1;
 	}
-	if (argc == optind)
+	if (argc == optind+1)
 	{
+		searchPattern = argv[argc-1];
 		text = readFile(stdin);
 	}
-	else
+	else if (argc == optind+2)
 	{
+		searchPattern = argv[argc-2];
 		char *filepath = argv[argc-1];
 		FILE *fp = fopen(filepath, "r");
 		if (fp == NULL)
@@ -108,10 +220,9 @@ int main(int argc, char *argv[])
 			return 0;
 		}
 	}
-	parseHtml(text);
-	free(tag);
-	free(attribute);
-	free(key);
+	struct filter_opts *options = parseFilterOpts(searchPattern);
+	filterHtml(text, options);
+	freeOpts(options);
 	free(text);
 	return 0;
 }
diff --git a/html.c b/html.c
@@ -438,7 +438,58 @@ void printHtml(struct tag *t)
 	}
 }
 
-void parseHtml(const char *text)
+struct tag *findTag(struct tag *tag, struct tag_list *list, struct filter_opts *opt)
+{
+	bool matchesTag = false;
+	bool matchesAttrKey = false;
+	bool matchesAttrKeyAndValue = false;
+	if (strcmp(tag->name, opt->tag) == 0)
+		matchesTag = true;
+	for (int i=0; i<tag->attrsLen; i++)
+	{
+		if (strcmp(tag->attrs[i]->name, opt->key) == 0)
+			matchesAttrKey = true;
+		if (matchesAttrKey)
+		{
+			if (strcmp(tag->attrs[i]->value, opt->attr) == 0)
+				matchesAttrKeyAndValue = true;
+		}
+	}
+	if (strlen(opt->tag) > 0 && strlen(opt->key) > 0 && strlen(opt->attr) > 0)
+	{
+		if (matchesTag && matchesAttrKeyAndValue)
+			return tag;
+	}
+	else if (strlen(opt->tag) > 0 && strlen(opt->key) > 0)
+	{
+		if (matchesTag && matchesAttrKey)
+			return tag;
+	}
+	else if (strlen(opt->tag) > 0)
+	{
+		if (matchesTag)
+			return tag;
+	}
+	else if (strlen(opt->key) > 0 && strlen(opt->attr) > 0)
+	{
+		if (matchesAttrKeyAndValue)
+			return tag;
+	}
+	else if (strlen(opt->key) > 0)
+	{
+		if (matchesAttrKey)
+			return tag;
+	}
+	for (int i=0; i<tag->childrenLen; i++)
+	{
+		struct tag *foundTag = findTag(tag->children[i], list, opt);
+		if (foundTag != NULL)
+			return foundTag;
+	}
+	return NULL;
+}
+
+void filterHtml(const char *text, struct filter_opts *opts)
 {
 	struct tag *rootTag;
 	struct tag_list *tagList = malloc(sizeof(struct tag_list));
@@ -449,10 +500,11 @@ void parseHtml(const char *text)
 		rootTag = parseTag(text+len, STATE_INNER_TEXT, tagList);
 	else
 		rootTag = parseTag(text, STATE_INNER_TEXT, tagList);
-	printHtml(rootTag);
-	/* printf("%s\n", rootTag->children[0]->children[0]->name);
-	printf("%s\n", rootTag->children[0]->children[0]->children[1]->name);
-	printf("%s\n", rootTag->children[0]->children[0]->children[1]->attrs[0]->value); */
+	struct tag *result = findTag(rootTag, tagList, opts);
+	if (result == NULL)
+		printError("No tag found.");
+	else
+		printf("result: %s\n", result->name);
 	freeTag(rootTag);
 	freeTagList(tagList);
 }
diff --git a/html.h b/html.h
@@ -19,6 +19,13 @@ const char *voidElements[] = {
 	"input", "link", "meta", "source", "track", "wbr"
 };
 
+struct filter_opts
+{
+	char *tag;
+	char *attr;
+	char *key;
+};
+
 struct attr
 {
 	char *name;
@@ -61,3 +68,10 @@ enum attr_value_syntax
 	AVS_APOSTROPHE,
 	AVS_UNQUOTED
 };
+
+/* enum search_type
+{
+	ST_NO,
+	ST_LIST,
+	ST_HIERARCHY
+}; */