cgmnlm

commit 48d0feed6d097c54662a7f231c7bc4704837f023
parent 33495e8dd86139cafade2888227e37b1572d18ea
Author: Drew DeVault <sir@cmpwn.com>
Date:   Sun, 20 Sep 2020 17:47:14 -0400

Initial pass on text/gemini parser

Diffstat:
M
configure
 | 
1
+
M
include/gmni.h
 | 
60
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
M
src/gmni.c
 | 
4
+++-
A
src/parser.c
 | 
144
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

4 files changed, 208 insertions(+), 1 deletion(-)
diff --git a/configure b/configure
@@ -7,6 +7,7 @@ gmni() {
 		src/client.c \
 		src/escape.c \
 		src/gmni.c \
+		src/parser.c \
 		src/url.c
 }
 
diff --git a/include/gmni.h b/include/gmni.h
@@ -103,4 +103,64 @@ char *gemini_input_url(const char *url, const char *input);
 // of the given Gemini status code.
 enum gemini_status_class gemini_response_class(enum gemini_status status);
 
+enum gemini_tok {
+	GEMINI_TEXT,
+	GEMINI_LINK,
+	GEMINI_PREFORMATTED,
+	GEMINI_HEADING,
+	GEMINI_LIST_ITEM,
+	GEMINI_QUOTE,
+};
+
+struct gemini_token {
+	enum gemini_tok token;
+
+	// The token field determines which of the union members is valid.
+	union {
+		char *text;
+
+		struct {
+			char *text;
+			char *url; // May be NULL
+		} link;
+
+		struct {
+			char *text;
+			char *alt_text; // May be NULL
+		} preformatted;
+
+		struct {
+			char *title;
+			int level; // 1, 2, or 3
+		} heading;
+
+		char *list_item;
+		char *quote_text;
+	};
+};
+
+struct gemini_parser {
+	BIO *f;
+	char *buf;
+	size_t bufsz;
+	size_t bufln;
+};
+
+// Initializes a text/gemini parser which reads from the specified BIO.
+void gemini_parser_init(struct gemini_parser *p, BIO *f);
+
+// Finishes this text/gemini parser and frees up its resources.
+void gemini_parser_finish(struct gemini_parser *p);
+
+// Reads the next token from a text/gemini file.
+// 
+// Returns 0 on success, 1 on EOF, and -1 on failure.
+//
+// Caller must call gemini_token_finish before exiting or re-using the token
+// parameter.
+int gemini_parser_next(struct gemini_parser *p, struct gemini_token *token);
+
+// Must be called after gemini_next to free up resources for the next token.
+void gemini_token_finish(struct gemini_token *token);
+
 #endif
diff --git a/src/gmni.c b/src/gmni.c
@@ -233,9 +233,11 @@ main(int argc, char *argv[])
 			printf("%d %s\n", resp.status, resp.meta);
 			/* fallthrough */
 		case OMIT_HEADERS:
-			if (resp.status / 10 != 2) {
+			if (gemini_response_class(resp.status) !=
+					GEMINI_STATUS_CLASS_SUCCESS) {
 				break;
 			}
+
 			char last;
 			char buf[BUFSIZ];
 			for (int n = 1; n > 0;) {
diff --git a/src/parser.c b/src/parser.c
@@ -0,0 +1,144 @@
+#include <assert.h>
+#include <ctype.h>
+#include <openssl/bio.h>
+#include <stddef.h>
+#include <stdlib.h>
+#include <string.h>
+#include "gmni.h"
+
+void
+gemini_parser_init(struct gemini_parser *p, BIO *f)
+{
+	p->f = f;
+	p->bufln = 0;
+	p->bufsz = BUFSIZ;
+	p->buf = malloc(p->bufsz + 1);
+	p->buf[0] = 0;
+	BIO_up_ref(p->f);
+}
+
+void
+gemini_parser_finish(struct gemini_parser *p)
+{
+	if (!p) {
+		return;
+	}
+	BIO_free(p->f);
+	free(p->buf);
+}
+
+int
+gemini_parser_next(struct gemini_parser *p, struct gemini_token *tok)
+{
+	memset(tok, 0, sizeof(*tok));
+
+	int eof = 0;
+	while (!strstr(p->buf, "\n")) {
+		if (p->bufln == p->bufsz) {
+			p->bufsz *= 2;
+			char *buf = realloc(p->buf, p->bufsz);
+			assert(buf);
+			p->buf = buf;
+		}
+
+		int n = BIO_read(p->f, &p->buf[p->bufln], p->bufsz - p->bufln);
+		if (n == -1) {
+			return -1;
+		} else if (n == 0) {
+			eof = 1;
+			break;
+		}
+		p->bufln += n;
+		p->buf[p->bufln] = 0;
+	}
+
+	// TODO: Collapse multi-line text for the user-agent to wrap
+	char *end;
+	if ((end = strstr(p->buf, "\n")) != NULL) {
+		*end = 0;
+	}
+
+	// TODO: Provide whitespace trimming helper function
+	if (strncmp(p->buf, "=>", 2) == 0) {
+		tok->token = GEMINI_LINK;
+		int i = 2;
+		while (p->buf[i] && isspace(p->buf[i])) ++i;
+		tok->link.url = &p->buf[i];
+
+		for (; p->buf[i]; ++i) {
+			if (isspace(p->buf[i])) {
+				p->buf[i++] = 0;
+				while (isspace(p->buf[i])) ++i;
+				if (p->buf[i]) {
+					tok->link.text = strdup(&p->buf[i]);
+				}
+				break;
+			}
+		}
+
+		tok->link.url = strdup(tok->link.url);
+	} else if (strncmp(p->buf, "```", 3) == 0) {
+		tok->token = GEMINI_PREFORMATTED; // TODO
+		tok->preformatted.text = strdup("<text>");
+		tok->preformatted.alt_text = strdup("<alt-text>");
+	} else if (p->buf[0] == '#') {
+		tok->token = GEMINI_HEADING;
+		int level = 1;
+		while (p->buf[level] == '#' && level < 3) {
+			++level;
+		}
+		tok->heading.level = level;
+		tok->heading.title = strdup(&p->buf[level]);
+	} else if (p->buf[0] == '*') {
+		tok->token = GEMINI_LIST_ITEM;
+		tok->list_item = strdup(&p->buf[1]);
+	} else if (p->buf[0] == '>') {
+		tok->token = GEMINI_QUOTE;
+		tok->quote_text = strdup(&p->buf[1]);
+	} else {
+		tok->token = GEMINI_TEXT;
+		tok->text = strdup(p->buf);
+	}
+
+	if (end && end + 1 < p->buf + p->bufln) {
+		size_t len = end - p->buf + 1;
+		memmove(p->buf, end + 1, p->bufln - len);
+		p->bufln -= len;
+	} else {
+		p->buf[0] = 0;
+		p->bufln = 0;
+	}
+
+	return eof;
+}
+
+void
+gemini_token_finish(struct gemini_token *tok)
+{
+	if (!tok) {
+		return;
+	}
+
+	switch (tok->token) {
+	case GEMINI_TEXT:
+		free(tok->text);
+		break;
+	case GEMINI_LINK:
+		free(tok->link.text);
+		free(tok->link.url);
+		break;
+	case GEMINI_PREFORMATTED:
+		free(tok->preformatted.text);
+		free(tok->preformatted.alt_text);
+		break;
+	case GEMINI_HEADING:
+		free(tok->heading.title);
+		break;
+	case GEMINI_LIST_ITEM:
+		free(tok->list_item);
+		break;
+	case GEMINI_QUOTE:
+		free(tok->quote_text);
+		break;
+	}
+}