htex

commit 445f2fb06c2b8cac70ed2e4496d78f9b5523e3ee
parent 18e781e88b49526a72b1405d9b0baeeb13e4fc58
Author: Robin <kroekerrobin@gmail.com>
Date:   Wed, 16 Aug 2023 11:28:12 +0200

Support --except argument and get attr value

Diffstat:
M
htex.c
 | 
45
++++++++++++++++++++++++++++-----------------
M
html.c
 | 
82
+++++++++++++++++++++++++++++++++++--------------------------------------------
M
html.h
 | 
3
++-
M
todo
 | 
1
-

4 files changed, 66 insertions(+), 65 deletions(-)
diff --git a/htex.c b/htex.c
@@ -128,37 +128,48 @@ void freeOpts(struct find_opts *opt)
 	free(opt);
 }
 
+enum output_type parseOutputArg(char *arg)
+{
+	if (arg == NULL)
+		return OUT_OUTER_HTML;
+	if (strcmp(arg, "innerhtml") == 0)
+		return OUT_INNER_HTML;
+	if (strcmp(arg, "innertext") == 0)
+		return OUT_INNER_TEXT;
+	if (strcmp(arg, "attr_value") == 0)
+		return OUT_ATTR_VALUE;
+	return -1;
+}
+
 int main(int argc, char *argv[])
 {
 	int o = 0;
 	int option_index = 0;
-	bool isInnerHtml = false;
-	bool isInnerText = false;
+	char *output = NULL;
 	bool isExcept = false;
 	char *text = NULL;
 	char *searchPattern = NULL;
 	static struct option long_options[] = {
-		{ "innerhtml", no_argument, 0, 'i' },
-		{ "innertext", no_argument, 0, 't' },
+		{ "output", required_argument, 0, 'o' },
 		{ "except", no_argument, 0, 'e' },
 		{ 0, 0, 0, 0 }
 	};
-	while ((o = getopt_long(argc, argv, "ite", long_options, &option_index)) != -1) {
+	while ((o = getopt_long(argc, argv, "o:e", long_options, &option_index)) != -1) {
 		switch(o) {
-			case 'i':
-				isInnerHtml = true;
-				break;
-			case 't':
-				isInnerText = true;
+			case 'o':
+				output = realloc(output, (strlen(optarg)+1) * sizeof(char));
+				strcpy(output, optarg);
 				break;
 			case 'e':
 				isExcept = true;
 				break;
 		}
 	}
-	if (isInnerHtml && isInnerText)
+	enum output_type out = parseOutputArg(output);
+	if (out == -1)
 	{
-		fprintf(stderr, "Provide either --innerhtml or --innertext.\n");
+		fprintf(stderr, "Provide a valid output type!\n");
+		free(output);
 		return -1;
 	}
 	if (argc == optind)
@@ -190,17 +201,17 @@ int main(int argc, char *argv[])
 		fclose(fp);
 		if (strlen(text) == 0)
 		{
-			printf("No data in file.\n");
+			fprintf(stderr, "No data in file.\n");
+			free(output);
+			free(text);
 			return 0;
 		}
 	}
 	struct find_opts *options = parseFilterOpts(searchPattern);
+	options->out = out;
 	options->isExcept = isExcept;
-	if (isInnerHtml)
-		options->out = OUT_INNER_HTML;
-	if (isInnerText)
-		options->out = OUT_INNER_TEXT;
 	filterHtml(text, options);
+	free(output);
 	freeOpts(options);
 	free(text);
 	return 0;
diff --git a/html.c b/html.c
@@ -676,48 +676,6 @@ void printHtml(struct tag *t, int indent)
 	}
 }
 
-void printTag(char *text, struct tag *t, enum output_type out, struct tag_list *foundTags)
-{
-	switch (out)
-	{
-		case OUT_INNER_HTML:
-			break;
-		case OUT_OUTER_HTML:
-			break;
-		case OUT_INNER_TEXT:
-			break;
-	}
-	/* unsigned int p = *(unsigned int *)t;
-	unsigned int cp;
-	bool isMatch = false;
-	for (int i=0; i<foundTags->len; i++)
-	{
-		cp = *(unsigned int *)foundTags->tags[i];
-		if (p == cp)
-			isMatch = true;
-	}
-	if (!isMatch)
-	{
-		char *trimmedText = NULL;
-		switch (out)
-		{
-			case OUT_INNER_HTML:
-				trimmedText = trim(getInnerHtml(text, t));
-				break;
-			case OUT_OUTER_HTML:
-				trimmedText = trim(getOuterHtml(text, t));
-				break;
-		}
-		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);
-	}
-}
-
 void printResult
 (
 	char *text,
@@ -728,7 +686,21 @@ void printResult
 {
 	if (opts->isExcept)
 	{
-		// printTag(text, rootTag, opts->out, foundTags);
+		bool isMatch = false;
+		for (int i=0; i<strlen(text); i++)
+		{
+			isMatch = false;
+			for (int k=0; k<foundTags->len; k++)
+			{
+				if (
+					foundTags->tags[k]->_outerHtmlBeginOffset <= i &&
+					foundTags->tags[k]->_outerHtmlEndOffset > i
+				)
+					isMatch = true;
+			}
+			if (!isMatch)
+				putchar(text[i]);
+		}
 	}
 	else
 	{
@@ -751,10 +723,28 @@ void printResult
 				case OUT_INNER_TEXT:
 					trimmedText = trim(foundTags->tags[i]->innerText);
 					break;
+				case OUT_ATTR_VALUE:
+					if (strlen(opts->key) > 0 && strlen(opts->tag) > 0)
+					{
+						for (int k=0; k<foundTags->tags[i]->attrsLen; k++)
+						{
+							if (strcmp(foundTags->tags[i]->attrs[k]->name, opts->key) == 0)
+								printf("%s\n", foundTags->tags[i]->attrs[k]->value);
+						}
+					}
+					else if (strlen(opts->tag) > 0)
+					{
+						for (int k=0; k<foundTags->tags[i]->attrsLen; k++)
+							printf("%s\n", foundTags->tags[i]->attrs[k]->value);
+					}
+					break;
+			}
+			if (trimmedText)
+			{
+				if (strlen(trimmedText) > 0)
+					printf("%s\n", trimmedText);
+				free(trimmedText);
 			}
-			if (strlen(trimmedText) > 0)
-				printf("%s\n", trimmedText);
-			free(trimmedText);
 		}
 	}
 }
diff --git a/html.h b/html.h
@@ -24,7 +24,8 @@ enum output_type
 {
 	OUT_INNER_HTML,
 	OUT_OUTER_HTML,
-	OUT_INNER_TEXT
+	OUT_INNER_TEXT,
+	OUT_ATTR_VALUE
 };
 
 struct find_opts
diff --git a/todo b/todo
@@ -1 +0,0 @@
-support --except argument