X-Git-Url: http://git.ozlabs.org/?p=ccan;a=blobdiff_plain;f=tools%2Fccanlint%2Ffile_analysis.c;h=8760630a7549e22a4d981c24fe69fab407d37043;hp=1341e57b11d3230780704ead95a0e123deb10042;hb=624871f35e94510b9924ba733f2b878ecf2cc6cc;hpb=650c775ff00cccd03fc84e7789a03c51d9839004

diff --git a/tools/ccanlint/file_analysis.c b/tools/ccanlint/file_analysis.c
index 1341e57b..8760630a 100644
--- a/tools/ccanlint/file_analysis.c
+++ b/tools/ccanlint/file_analysis.c
@@ -1,8 +1,17 @@
+#include "config.h"
 #include "ccanlint.h"
-#include "get_file_lines.h"
-#include <talloc/talloc.h>
-#include <string/string.h>
-#include <noerr/noerr.h>
+#include <ccan/talloc/talloc.h>
+#include <ccan/str/str.h>
+#include <ccan/str_talloc/str_talloc.h>
+#include <ccan/talloc_link/talloc_link.h>
+#include <ccan/hash/hash.h>
+#include <ccan/htable/htable_type.h>
+#include <ccan/grab_file/grab_file.h>
+#include <ccan/noerr/noerr.h>
+#include <ccan/foreach/foreach.h>
+#include <ccan/asort/asort.h>
+#include <ccan/array_size/array_size.h>
+#include "../tools.h"
 #include <unistd.h>
 #include <sys/types.h>
 #include <sys/stat.h>
@@ -10,140 +19,378 @@
 #include <err.h>
 #include <errno.h>
 #include <dirent.h>
+#include <ctype.h>
+#include <stdarg.h>
+#include <assert.h>
 
-char **get_ccan_file_lines(struct ccan_file *f)
+struct list_head *get_ccan_file_docs(struct ccan_file *f)
 {
-	if (!f->lines)
-		f->lines = get_file_lines(f, f->name, &f->num_lines);
-	return f->lines;
+	if (!f->doc_sections) {
+		get_ccan_file_lines(f);
+		f->doc_sections = extract_doc_sections(f->lines);
+	}
+	return f->doc_sections;
 }
 
-static void add_files(struct manifest *m, const char *dir)
+
+/**
+ * remove_comments - strip comments from a line, return copy.
+ * @line: line to copy
+ * @in_comment: are we already within a comment (from prev line).
+ * @unterminated: are we still in a comment for next line.
+ */
+static char *remove_comments(const char *line, bool in_comment,
+			     bool *unterminated)
 {
-	DIR *d;
-	struct dirent *ent;
+	char *p, *ret = talloc_array(line, char, strlen(line) + 1);
 
-	if (dir[0])
-		d = opendir(dir);
-	else
-		d = opendir(".");
-	if (!d)
-		err(1, "Opening directory %s", dir[0] ? dir : ".");
+	p = ret;
+	for (;;) {
+		if (!in_comment) {
+			/* Find first comment. */
+			const char *old_comment = strstr(line, "/*");
+			const char *new_comment = strstr(line, "//");
+			const char *comment;
 
-	while ((ent = readdir(d)) != NULL) {
-		struct stat st;
-		struct ccan_file *f;
-		struct list_head *dest;
-		bool is_c_src;
+			if (new_comment && old_comment)
+				comment = new_comment < old_comment
+					? new_comment : old_comment;
+			else if (old_comment)
+				comment = old_comment;
+			else if (new_comment)
+				comment = new_comment;
+			else {
+				/* Nothing more. */
+				strcpy(p, line);
+				*unterminated = false;
+				break;
+			}
 
-		if (ent->d_name[0] == '.')
-			continue;
+			/* Copy up to comment. */
+			memcpy(p, line, comment - line);
+			p += comment - line;
+			line += comment - line + 2;
 
-		f = talloc(m, struct ccan_file);
-		f->lines = NULL;
-		f->name = talloc_asprintf(f, "%s%s", dir, ent->d_name);
-		if (lstat(f->name, &st) != 0)
-			err(1, "lstat %s", f->name);
+			if (comment == new_comment) {
+				/* We're done: goes to EOL. */
+				p[0] = '\0';
+				*unterminated = false;
+				break;
+			}
+			in_comment = true;
+		}
 
-		if (S_ISDIR(st.st_mode)) {
-			f->name = talloc_append_string(f->name, "/");
-			add_files(m, f->name);
-			continue;
+		if (in_comment) {
+			const char *end = strstr(line, "*/");
+			if (!end) {
+				*unterminated = true;
+				p[0] = '\0';
+				break;
+			}
+			line = end+2;
+			in_comment = false;
 		}
-		if (!S_ISREG(st.st_mode)) {
-			talloc_free(f);
+	}
+	return ret;
+}
+
+static bool is_empty(const char *line)
+{
+	return strspn(line, " \r\t") == strlen(line);
+}
+
+static bool continues(const char *line)
+{
+	/* Technically, any odd number of these.  But who cares? */
+	return strends(line, "\\");
+}
+
+static bool parse_hash_if(struct pp_conditions *cond, const char **line)
+{
+	bool brackets, defined;
+
+	cond->inverse = get_token(line, "!");
+	defined = get_token(line, "defined");
+	brackets = get_token(line, "(");
+	cond->symbol = get_symbol_token(cond, line);
+	if (!cond->symbol)
+		return false;
+	if (brackets && !get_token(line, ")"))
+		return false;
+	if (!defined)
+		cond->type = PP_COND_IF;
+
+	/* FIXME: We just chain them, ignoring operators. */
+	if (get_token(line, "||") || get_token(line, "&&")) {
+		struct pp_conditions *sub = talloc(cond, struct pp_conditions);
+
+		sub->parent = cond->parent;
+		sub->type = PP_COND_IFDEF;
+		if (parse_hash_if(sub, line))
+			cond->parent = sub;
+	}
+
+	return true;
+}
+
+/* FIXME: Get serious! */
+static struct pp_conditions *analyze_directive(struct ccan_file *f,
+					       const char *line,
+					       struct pp_conditions *parent)
+{
+	struct pp_conditions *cond = talloc(f, struct pp_conditions);
+	bool unused;
+
+	line = remove_comments(line, false, &unused);
+
+	cond->parent = parent;
+	cond->type = PP_COND_IFDEF;
+
+	if (!get_token(&line, "#"))
+		abort();
+
+	if (get_token(&line, "if")) {
+		if (!parse_hash_if(cond, &line))
+			goto unknown;
+	} else if (get_token(&line, "elif")) {
+		/* Malformed? */
+		if (!parent)
+			return NULL;
+		cond->parent = parent->parent;
+		/* FIXME: Not quite true.  This implies !parent, but we don't
+		 * do multiple conditionals yet. */
+		if (!parse_hash_if(cond, &line))
+			goto unknown;
+	} else if (get_token(&line, "ifdef")) {
+		bool brackets;
+		cond->inverse = false;
+		brackets = get_token(&line, "(");
+		cond->symbol = get_symbol_token(cond, &line);
+		if (!cond->symbol)
+			goto unknown;
+		if (brackets && !get_token(&line, ")"))
+			goto unknown;
+	} else if (get_token(&line, "ifndef")) {
+		bool brackets;
+		cond->inverse = true;
+		brackets = get_token(&line, "(");
+		cond->symbol = get_symbol_token(cond, &line);
+		if (!cond->symbol)
+			goto unknown;
+		if (brackets && !get_token(&line, ")"))
+			goto unknown;
+	} else if (get_token(&line, "else")) {
+		/* Malformed? */
+		if (!parent)
+			return NULL;
+
+		*cond = *parent;
+		cond->inverse = !cond->inverse;
+		return cond;
+	} else if (get_token(&line, "endif")) {
+		talloc_free(cond);
+		/* Malformed? */
+		if (!parent)
+			return NULL;
+		/* Back up one! */
+		return parent->parent;
+	} else {
+		/* Not a conditional. */
+		talloc_free(cond);
+		return parent;
+	}
+
+	if (!is_empty(line))
+		goto unknown;
+	return cond;
+
+unknown:
+	cond->type = PP_COND_UNKNOWN;
+	return cond;
+}
+
+/* This parser is rough, but OK if code is reasonably neat. */
+struct line_info *get_ccan_line_info(struct ccan_file *f)
+{
+	bool continued = false, in_comment = false;
+	struct pp_conditions *cond = NULL;
+	unsigned int i;
+
+	if (f->line_info)
+		return f->line_info;
+
+	get_ccan_file_lines(f);
+	f->line_info = talloc_array(f->lines, struct line_info, f->num_lines);
+
+	for (i = 0; i < f->num_lines; continued = continues(f->lines[i++])) {
+		char *p;
+		bool still_doc_line;
+
+		/* Current conditions apply to this line. */
+		f->line_info[i].cond = cond;
+		f->line_info[i].continued = continued;
+
+		if (continued) {
+			/* Same as last line. */
+			f->line_info[i].type = f->line_info[i-1].type;
+			/* Update in_comment. */
+			remove_comments(f->lines[i], in_comment, &in_comment);
 			continue;
 		}
 
-		if (streq(f->name, "_info.c")) {
-			m->info_file = f;
+		/* Preprocessor directive? */
+		if (!in_comment
+		    && f->lines[i][strspn(f->lines[i], " \t")] == '#') {
+			f->line_info[i].type = PREPROC_LINE;
+			cond = analyze_directive(f, f->lines[i], cond);
 			continue;
 		}
 
-		is_c_src = strends(f->name, ".c");
-		if (!is_c_src && !strends(f->name, ".h"))
-			dest = &m->other_files;
-		else if (!strchr(f->name, '/')) {
-			if (is_c_src)
-				dest = &m->c_files;
+		still_doc_line = (in_comment
+				  && f->line_info[i-1].type == DOC_LINE);
+
+		p = remove_comments(f->lines[i], in_comment, &in_comment);
+		if (is_empty(p)) {
+			if (strstarts(f->lines[i], "/**") || still_doc_line)
+				f->line_info[i].type = DOC_LINE;
 			else
-				dest = &m->h_files;
-		} else if (strstarts(f->name, "test/")) {
-			if (is_c_src) {
-				if (strstarts(f->name, "test/run"))
-					dest = &m->run_tests;
-				else if (strstarts(f->name, "test/compile_ok"))
-					dest = &m->compile_ok_tests;
-				else if (strstarts(f->name, "test/compile_fail"))
-					dest = &m->compile_fail_tests;
-				else
-					dest = &m->other_test_files;
-			} else
-				dest = &m->other_test_files;
+				f->line_info[i].type = COMMENT_LINE;
 		} else
-			dest = &m->other_files;
-
-		list_add(dest, &f->list);
+			f->line_info[i].type = CODE_LINE;
+		talloc_free(p);
 	}
-	closedir(d);
+	return f->line_info;
+}
+
+struct symbol {
+	struct list_node list;
+	const char *name;
+	const unsigned int *value;
+};
+
+static struct symbol *find_symbol(struct list_head *syms, const char *sym)
+{
+	struct symbol *i;
+
+	list_for_each(syms, i, list)
+		if (streq(sym, i->name))
+			return i;
+	return NULL;
 }
 
-char *report_on_lines(struct list_head *files,
-		      char *(*report)(const char *),
-		      char *sofar)
+static enum line_compiled get_pp(struct pp_conditions *cond,
+				 struct list_head *syms)
 {
-	struct ccan_file *f;
+	struct symbol *sym;
+	unsigned int val;
+	enum line_compiled parent, ret;
 
-	list_for_each(files, f, list) {
-		unsigned int i;
-		char **lines = get_ccan_file_lines(f);
+	/* No conditions?  Easy. */
+	if (!cond)
+		return COMPILED;
 
-		for (i = 0; i < f->num_lines; i++) {
-			char *r = report(lines[i]);
-			if (!r)
-				continue;
+	/* Check we get here at all. */
+	parent = get_pp(cond->parent, syms);
+	if (parent == NOT_COMPILED)
+		return NOT_COMPILED;
 
-			sofar = talloc_asprintf_append(sofar,
-						       "%s:%u:%s\n",
-						       f->name, i+1, r);
-			talloc_free(r);
-		}
+	if (cond->type == PP_COND_UNKNOWN)
+		return MAYBE_COMPILED;
+
+	sym = find_symbol(syms, cond->symbol);
+	if (!sym)
+		return MAYBE_COMPILED;
+
+	switch (cond->type) {
+	case PP_COND_IF:
+		/* Undefined is 0. */
+		val = sym->value ? *sym->value : 0;
+		if (!val == cond->inverse)
+			ret = COMPILED;
+		else
+			ret = NOT_COMPILED;
+		break;
+
+	case PP_COND_IFDEF:
+		if (cond->inverse == !sym->value)
+			ret = COMPILED;
+		else
+			ret = NOT_COMPILED;
+		break;
+
+	default:
+		abort();
 	}
-	return sofar;
+
+	/* If parent didn't know, NO == NO, but YES == MAYBE. */
+	if (parent == MAYBE_COMPILED && ret == COMPILED)
+		ret = MAYBE_COMPILED;
+	return ret;
 }
 
-struct manifest *get_manifest(void)
+static void add_symbol(struct list_head *head,
+		       const char *symbol, const unsigned int *value)
+{
+	struct symbol *sym = talloc(head, struct symbol);
+	sym->name = symbol;
+	sym->value = value;
+	list_add(head, &sym->list);
+}
+	
+enum line_compiled get_ccan_line_pp(struct pp_conditions *cond,
+				    const char *symbol,
+				    const unsigned int *value,
+				    ...)
 {
-	struct manifest *m = talloc(NULL, struct manifest);
-	unsigned int len;
-
-	m->info_file = NULL;
-	list_head_init(&m->c_files);
-	list_head_init(&m->h_files);
-	list_head_init(&m->run_tests);
-	list_head_init(&m->compile_ok_tests);
-	list_head_init(&m->compile_fail_tests);
-	list_head_init(&m->other_test_files);
-	list_head_init(&m->other_files);
-
-	/* *This* is why people hate C. */
-	len = 32;
-	m->basename = talloc_array(m, char, len);
-	while (!getcwd(m->basename, len)) {
-		if (errno != ERANGE)
-			err(1, "Getting current directory");
-		m->basename = talloc_realloc(m, m->basename, char, len *= 2);
+	enum line_compiled ret;
+	struct list_head *head;
+	va_list ap;
+
+	head = talloc(NULL, struct list_head);
+	list_head_init(head);
+
+	va_start(ap, value);
+	add_symbol(head, symbol, value);
+
+	while ((symbol = va_arg(ap, const char *)) != NULL) {
+		value = va_arg(ap, const unsigned int *);
+		add_symbol(head, symbol, value);
 	}
+	ret = get_pp(cond, head);
+	talloc_free(head);
+	return ret;
+}
+
+void score_file_error(struct score *score, struct ccan_file *f, unsigned line,
+		      const char *errorfmt, ...)
+{
+	va_list ap;
+
+	struct file_error *fe = talloc(score, struct file_error);
+	fe->file = f;
+	fe->line = line;
+	list_add_tail(&score->per_file_errors, &fe->list);
 
-	len = strlen(m->basename);
-	while (len && m->basename[len-1] == '/')
-		m->basename[--len] = '\0';
+	if (!score->error)
+		score->error = talloc_strdup(score, "");
+	
+	if (verbose < 2 && strcount(score->error, "\n") > 5)
+		return;
+
+	if (line)
+		score->error = talloc_asprintf_append(score->error,
+						      "%s:%u:",
+						      f->fullname, line);
+	else
+		score->error = talloc_asprintf_append(score->error,
+						      "%s:", f->fullname);
 
-	m->basename = strrchr(m->basename, '/');
-	if (!m->basename)
-		errx(1, "I don't expect to be run from the root directory");
-	m->basename++;
+	va_start(ap, errorfmt);
+	score->error = talloc_vasprintf_append(score->error, errorfmt, ap);
+	va_end(ap);
+	score->error = talloc_append_string(score->error, "\n");
 
-	add_files(m, "");
-	return m;
+	if (verbose < 2 && strcount(score->error, "\n") > 5)
+		score->error = talloc_append_string(score->error,
+				    "... more (use -vv to see them all)\n");
 }