From 09dc36d26100236bc58969fda149f2c8b55bb1cb Mon Sep 17 00:00:00 2001 From: harshitaajoshi Date: Sun, 27 Sep 2026 18:16:16 -0500 Subject: [PATCH] fix(extract): parse PHP past inline HTML so symbols after ?> reach the graph (#2000) Signed-off-by: harshitaajoshi --- Makefile.cbm | 1 + internal/cbm/cbm.c | 14 ++ internal/cbm/cbm.h | 4 + internal/cbm/lsp/php_lsp.c | 4 + internal/cbm/php_inline_html.c | 317 +++++++++++++++++++++++++++++++++ tests/test_extraction.c | 128 +++++++++++++ tests/test_pipeline.c | 60 +++++++ 7 files changed, 528 insertions(+) create mode 100644 internal/cbm/php_inline_html.c diff --git a/Makefile.cbm b/Makefile.cbm index 6d19f430a..fec175f67 100644 --- a/Makefile.cbm +++ b/Makefile.cbm @@ -398,6 +398,7 @@ EXTRACTION_SRCS = \ $(CBM_DIR)/route_path.c \ $(CBM_DIR)/openapi.c \ $(CBM_DIR)/sql_values.c \ + $(CBM_DIR)/php_inline_html.c \ $(CBM_DIR)/helpers.c \ $(CBM_DIR)/cpp_branch_views.c \ $(CBM_DIR)/result_compact.c \ diff --git a/internal/cbm/cbm.c b/internal/cbm/cbm.c index 23b871dbc..4eb030ae0 100644 --- a/internal/cbm/cbm.c +++ b/internal/cbm/cbm.c @@ -3313,6 +3313,18 @@ CBMFileResult *cbm_extract_file(const char *source, int source_len, CBMLanguage return r; } +/* The bytes the grammar parses. PHP that leaves PHP mode (`?> markup ` (#2000). Same length and line breaks as the input; returns +// `source` itself when nothing needs rewriting. Defined in php_inline_html.c. +const char *cbm_php_mask_inline_html(CBMArena *arena, const char *source, int source_len); void cbm_extract_imports(CBMExtractCtx *ctx); void cbm_extract_usages(CBMExtractCtx *ctx); void cbm_extract_semantic(CBMExtractCtx *ctx); diff --git a/internal/cbm/lsp/php_lsp.c b/internal/cbm/lsp/php_lsp.c index 7aae329e8..85118f671 100644 --- a/internal/cbm/lsp/php_lsp.c +++ b/internal/cbm/lsp/php_lsp.c @@ -20,6 +20,7 @@ #include "php_lsp.h" #include "lsp_node_iter.h" #include "lsp_work.h" +#include "../cbm.h" #include "../helpers.h" #include "../../../src/foundation/hash_table.h" #include @@ -4795,6 +4796,9 @@ void cbm_run_php_lsp_cross_with_bindings(CBMArena *arena, const char *source, in const CBMPHPImportBindings *bindings) { if (!arena || !source || source_len <= 0 || !out) return; + /* The retained source is the file on disk; read it the way the per-file + * extract did, so the cached tree and node text agree (#2000). */ + source = cbm_php_mask_inline_html(arena, source, source_len); TSParser *parser = NULL; TSTree *tree = cached_tree; diff --git a/internal/cbm/php_inline_html.c b/internal/cbm/php_inline_html.c new file mode 100644 index 000000000..df5652d54 --- /dev/null +++ b/internal/cbm/php_inline_html.c @@ -0,0 +1,317 @@ +// php_inline_html.c — let the php_only grammar read a file that leaves PHP mode. +// +// A PHP file starts in inline-HTML mode and may switch in and out of PHP any +// number of times: `?>
...
` or for the +// text that follows it, so every byte after the first close tag used to land in +// an ERROR region and the declarations there never reached the graph (#2000). +// +// Instead of parsing the markup, blank it. The rewritten buffer has the same +// length as the original and keeps every line break, so byte offsets, rows and +// columns are unchanged and node text inside PHP code is byte-identical: +// +// - markup before the first open tag and between tags becomes spaces +// - the first open tag stays, so the tree keeps its php_tag +// - each later `` becomes `; `, the statement terminator PHP itself treats it as +// +// Only code and `//` / `#` comments can close PHP mode. Strings, heredoc, +// nowdoc and block comments are skipped, `#[` is an attribute, and everything +// after __halt_compiler is left alone, as PHP stops reading there. + +#include "cbm.h" +#include "foundation/arena.h" +#include "foundation/constants.h" +#include +#include +#include + +enum { + PHP_TAG_OPEN_LEN = 5, /* = 0 && i <= len && (size_t)(len - i) >= n && memcmp(s + i, lit, n) == 0; +} + +static bool php_is_ident_char(unsigned char c) { + return c == '_' || (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') || (c >= '0' && c <= '9') || + c >= PHP_FIRST_NON_ASCII; +} + +static bool php_is_ident_start(unsigned char c) { + return php_is_ident_char(c) && !(c >= '0' && c <= '9'); +} + +static bool php_is_space(unsigned char c) { + return c == ' ' || c == '\t' || c == '\n' || c == '\r' || c == '\f' || c == '\v'; +} + +static bool php_is_blank(char c) { + return c == ' ' || c == '\t'; +} + +/* Case-insensitive match of the lowercase `word` at s[i]. */ +static bool php_ieq(const char *s, int len, int i, const char *word) { + int n = (int)strlen(word); + if (i + n > len) { + return false; + } + for (int k = 0; k < n; k++) { + char c = s[i + k]; + if (c >= 'A' && c <= 'Z') { + c = (char)(c - 'A' + 'a'); + } + if (c != word[k]) { + return false; + } + } + return true; +} + +/* Length of the open tag at i, or 0. `` prolog in markup stays markup. */ +static int php_open_tag_len(const char *s, int len, int i) { + if (!php_at(s, len, i, "")) { + return true; + } + } + return false; +} + +/* Blank [from, to) in place, keeping line breaks. */ +static void php_blank(char *out, int from, int to) { + for (int k = from; k < to; k++) { + if (out[k] != '\n' && out[k] != '\r') { + out[k] = ' '; + } + } +} + +/* Index just past the quoted string opened at i. */ +static int php_skip_quoted(const char *s, int len, int i) { + char quote = s[i]; + i += SKIP_ONE; + while (i < len && s[i] != quote) { + i += (s[i] == '\\') ? PAIR_LEN : SKIP_ONE; + } + return i < len ? i + SKIP_ONE : len; +} + +/* Index just past the block comment opened at i. */ +static int php_skip_block_comment(const char *s, int len, int i) { + for (int k = i + PAIR_LEN; k < len; k++) { + if (php_at(s, len, k, "*/")) { + return k + PAIR_LEN; + } + } + return len; +} + +/* Index of the line break or `?>` that ends the one-line comment at i. */ +static int php_skip_line_comment(const char *s, int len, int i) { + while (i < len && s[i] != '\n' && !php_at(s, len, i, "?>")) { + i++; + } + return i; +} + +/* Parse the heredoc/nowdoc opener `<<= len || s[p] != quote) { + return PHP_NOT_FOUND; + } + p++; + } + if (p < len && s[p] == '\r') { + p++; + } + return (p < len && s[p] == '\n') ? p : PHP_NOT_FOUND; +} + +/* True when the line starting at p closes the heredoc. PHP 7.3+ allows the + * closing label to be indented and followed by any non-identifier byte. */ +static bool php_heredoc_closes(const char *s, int len, int p, int label, int label_len, int *end) { + while (p < len && php_is_blank(s[p])) { + p++; + } + if (len - p < label_len || memcmp(s + p, s + label, (size_t)label_len) != 0) { + return false; + } + int after = p + label_len; + if (after < len && php_is_ident_char((unsigned char)s[after])) { + return false; + } + *end = after; + return true; +} + +/* Index just past the heredoc/nowdoc opened by the `<<<` at i. */ +static int php_skip_heredoc(const char *s, int len, int i) { + int label = 0; + int label_len = 0; + int nl = php_heredoc_opener(s, len, i, &label, &label_len); + if (nl < 0) { + return i + PHP_HEREDOC_LEN; + } + while (nl < len) { + int line = nl + SKIP_ONE; + int end = 0; + if (php_heredoc_closes(s, len, line, label, label_len, &end)) { + return end; + } + const char *next = memchr(s + line, '\n', (size_t)(len - line)); + if (!next) { + return len; + } + nl = (int)(next - s); + } + return len; +} + +/* Index just past the identifier at i; sets *halt when it is __halt_compiler + * (a keyword, so not a `$variable` of that name). */ +static int php_skip_ident(const char *s, int len, int i, bool *halt) { + int start = i; + while (i < len && php_is_ident_char((unsigned char)s[i])) { + i++; + } + bool is_var = start > 0 && s[start - SKIP_ONE] == '$'; + *halt = + !is_var && i - start == PHP_HALT_COMPILER_LEN && php_ieq(s, len, start, PHP_HALT_COMPILER); + return i; +} + +/* Scan PHP code from i to the next close tag. Returns the index of that `?>`, + * len when the file ends in PHP mode, or PHP_NOT_FOUND at __halt_compiler. */ +static int php_scan_code(const char *s, int len, int i) { + while (i < len) { + unsigned char c = (unsigned char)s[i]; + if (php_at(s, len, i, "?>")) { + return i; + } + if (c == '\'' || c == '"' || c == '`') { + i = php_skip_quoted(s, len, i); + } else if (php_at(s, len, i, "/*")) { + i = php_skip_block_comment(s, len, i); + } else if (php_at(s, len, i, "//") || (c == '#' && !php_at(s, len, i, "#["))) { + i = php_skip_line_comment(s, len, i); + } else if (php_at(s, len, i, "<<<")) { + i = php_skip_heredoc(s, len, i); + } else if (php_is_ident_start(c)) { + bool halt = false; + i = php_skip_ident(s, len, i, &halt); + if (halt) { + return PHP_NOT_FOUND; + } + } else { + i++; + } + } + return len; +} + +const char *cbm_php_mask_inline_html(CBMArena *arena, const char *source, int source_len) { + if (!arena || !source || source_len <= 0) { + return source; + } + /* Fast path: a file that opens PHP at its first byte and never writes a + * close tag is exactly what the grammar already accepts. */ + if (php_open_tag_len(source, source_len, 0) > 0 && !php_has_close_tag(source, source_len)) { + return source; + } + + char *out = (char *)cbm_arena_alloc(arena, (size_t)source_len + SKIP_ONE); + if (!out) { + return source; + } + memcpy(out, source, (size_t)source_len); + out[source_len] = '\0'; + + bool changed = false; + bool seen_open = false; + int i = 0; + while (i < source_len) { + /* Inline-HTML mode: blank up to the next open tag. */ + int html_start = i; + int tag = 0; + while (i < source_len && (tag = php_open_tag_len(source, source_len, i)) == 0) { + i++; + } + if (i > html_start) { + php_blank(out, html_start, i); + changed = true; + } + if (i >= source_len) { + break; + } + if (seen_open) { + php_blank(out, i, i + tag); + changed = true; + } + seen_open = true; + + /* PHP mode, up to the close tag. */ + int close = php_scan_code(source, source_len, i + tag); + if (close < 0 || close >= source_len) { + break; + } + out[close] = ';'; + out[close + SKIP_ONE] = ' '; + changed = true; + i = close + PAIR_LEN; + } + return changed ? out : source; +} diff --git a/tests/test_extraction.c b/tests/test_extraction.c index 7bfda63fc..c8d47ca52 100644 --- a/tests/test_extraction.c +++ b/tests/test_extraction.c @@ -833,6 +833,130 @@ TEST(php_function) { PASS(); } +/* #2000: inline HTML between `?>` and `` fell into an + * ERROR region and never reached the graph. */ +static const CBMCall *find_call_named(CBMFileResult *r, const char *callee) { + for (int i = 0; i < r->calls.count; i++) { + if (r->calls.items[i].callee_name && strcmp(r->calls.items[i].callee_name, callee) == 0) + return &r->calls.items[i]; + } + return NULL; +} + +static int php_def_line(CBMFileResult *r, const char *name) { + for (int i = 0; i < r->defs.count; i++) { + if (strcmp(r->defs.items[i].name, name) == 0) + return (int)r->defs.items[i].start_line; + } + return -1; +} + +TEST(php_inline_html_tail_reaches_graph_issue2000) { + CBMFileResult *r = extract("\n" + "
\n" + "

plain markup after the closing tag

\n" + "
\n" + "has_error); + ASSERT(has_def(r, "Function", "gamma")); + ASSERT(has_def(r, "Function", "delta")); + /* Lines are in original-file coordinates. */ + ASSERT_EQ(php_def_line(r, "gamma"), 2); + ASSERT_EQ(php_def_line(r, "delta"), 8); + const CBMCall *c = find_call_named(r, "gamma"); + ASSERT_NOT_NULL(c); + ASSERT_NOT_NULL(strstr(c->enclosing_func_qn, "delta")); + ASSERT_FALSE(r->parse_incomplete); + cbm_free_result(r); + PASS(); +} + +/* View-file shapes: echo tags keep their calls, and control structures that + * are opened in one PHP block and closed in a later one still parse. */ +TEST(php_inline_html_template_shapes_issue2000) { + CBMFileResult *r = extract("\n" + "<?= page_title($page) ?>\n" + "\n" + "\n" + "

Hello name?>

\n" + "\n" + " log in\n" + "\n" + "\n" + "
    • \n" + "\n" + "\n" + "\n", + CBM_LANG_PHP, "t", "view.php"); + ASSERT_NOT_NULL(r); + ASSERT_FALSE(r->has_error); + ASSERT(has_call(r, "page_title")); + ASSERT(has_call(r, "render_item")); + ASSERT(has_def(r, "Function", "footer_links")); + ASSERT_EQ(php_def_line(r, "footer_links"), 12); + ASSERT(has_call(r, "nav_links")); + ASSERT_FALSE(r->parse_incomplete); + cbm_free_result(r); + PASS(); +} + +/* `?>` only leaves PHP mode from code or a one-line comment. Inside strings, + * heredoc, nowdoc and block comments it is text; `#[` is an attribute, not a + * comment; ` quoted';\n" + "$b = \"double ?> quoted \\\" still\";\n" + "$c = << body\n" + " EOT;\n" + "$d = <<<'NOW'\n" + "nowdoc ?> body\n" + "NOW;\n" + "/* block ?> comment */\n" + "#[Pure]\n" + "function one() { return 1; }\n" + "// line comment ends here ?>\n" + "\n" + "has_error); + ASSERT(has_def(r, "Function", "one")); + ASSERT(has_def(r, "Function", "two")); + ASSERT_EQ(php_def_line(r, "two"), 16); + ASSERT(has_call(r, "one")); + ASSERT_FALSE(r->parse_incomplete); + cbm_free_result(r); + PASS(); +} + +/* A trailing `?>` (common at the end of older class files) is not a partial + * parse, and CRLF line endings keep their line numbers. */ +TEST(php_inline_html_trailing_close_tag_crlf_issue2000) { + CBMFileResult *r = extract("\r\n", + CBM_LANG_PHP, "t", "Repo.php"); + ASSERT_NOT_NULL(r); + ASSERT_FALSE(r->has_error); + ASSERT(has_def(r, "Class", "Repo")); + ASSERT(has_def(r, "Method", "find")); + ASSERT_EQ(php_def_line(r, "find"), 3); + ASSERT_FALSE(r->parse_incomplete); + cbm_free_result(r); + PASS(); +} + /* --- Ruby --- */ TEST(ruby_class) { CBMFileResult *r = extract("class Animal\n def initialize(name)\n @name = name\n end\n " @@ -13125,6 +13249,10 @@ SUITE(extraction) { RUN_TEST(python_class_base_extracted_bare); RUN_TEST(php_class); RUN_TEST(php_function); + RUN_TEST(php_inline_html_tail_reaches_graph_issue2000); + RUN_TEST(php_inline_html_template_shapes_issue2000); + RUN_TEST(php_inline_html_close_tag_in_literals_issue2000); + RUN_TEST(php_inline_html_trailing_close_tag_crlf_issue2000); RUN_TEST(ruby_class); RUN_TEST(ruby_module); RUN_TEST(csharp_class); diff --git a/tests/test_pipeline.c b/tests/test_pipeline.c index d9b327226..94a281a07 100644 --- a/tests/test_pipeline.c +++ b/tests/test_pipeline.c @@ -1376,6 +1376,65 @@ TEST(pipeline_nix_scoped_binding_calls_resolve) { PASS(); } +/* #2000, end to end: functions declared after an inline-HTML block reach the + * store with their CALLS edges, in the same file and across files. */ +TEST(pipeline_php_inline_html_calls_resolve_issue2000) { + if (setup_test_repo() != 0) { + FAIL("failed to create temp dir"); + } + + static const struct { + const char *name; + const char *body; + } files[] = { + {"inline_html.php", "\n" + "
    \n" + "

    plain markup after the closing tag

    \n" + "
    \n" + "\n" + "\n" + "

    \n" + "