Add a C API for on_end_tag

c-api/include/lol_html.h

@@ -23,6 +23,7 @@ typedef struct lol_html_HtmlRewriterBuilder lol_html_rewriter_builder_t;
 typedef struct lol_html_HtmlRewriter lol_html_rewriter_t;
 typedef struct lol_html_Doctype lol_html_doctype_t;
 typedef struct lol_html_DocumentEnd lol_html_doc_end_t;
+typedef struct lol_html_EndTag lol_html_end_tag_t;
 typedef struct lol_html_Comment lol_html_comment_t;
 typedef struct lol_html_TextChunk lol_html_text_chunk_t;
 typedef struct lol_html_Element lol_html_element_t;
@@ -110,6 +111,11 @@ typedef lol_html_rewriter_directive_t (*lol_html_doc_end_handler_t)(
     void *user_data
 );
 
+typedef lol_html_rewriter_directive_t (*lol_html_end_tag_handler_t)(
+    lol_html_end_tag_t *end_tag,
+    void *user_data
+);
+
 // Selector
 //---------------------------------------------------------------------
 
@@ -671,6 +677,69 @@ void lol_html_element_user_data_set(
 // Returns user data attached to the text chunk.
 void *lol_html_element_user_data_get(const lol_html_element_t *element);
 
+// Adds content handlers to the builder for the end tag of the given element.
+//
+// Subsequent calls to the method on the same element replace the previous handler.
+//
+// The handler can optionally have associated user data which will be
+// passed to the handler on each invocation along with the rewritable
+// unit argument.
+//
+// If the handler returns LOL_HTML_STOP directive then rewriting
+// stops immediately and `write()` or `end()` of the rewriter methods
+// return an error code.
+//
+// Returns 0 in case of success and -1 otherwise. The actual error message
+// can be obtained using `lol_html_take_last_error` function.
+//
+// WARNING: Pointers passed to handlers are valid only during the
+// handler execution. So they should never be leaked outside of handlers.
+int lol_html_element_on_end_tag(lol_html_element_t* element, lol_html_end_tag_handler_t end_tag_handler, void* user_data);
+
+// Inserts the content string before the element's end tag either as raw text or as HTML.
+//
+// Content should be a valid UTF8-string.
+//
+// Returns 0 in case of success and -1 otherwise. The actual error message
+// can be obtained using `lol_html_take_last_error` function.
+int lol_html_end_tag_before(
+    lol_html_end_tag_t *end_tag,
+    const char *content,
+    size_t content_len,
+    bool is_html
+);
+
+// Inserts the content string right after the element's end tag as raw text or as HTML.
+//
+// Content should be a valid UTF8-string.
+//
+// Returns 0 in case of success and -1 otherwise. The actual error message
+// can be obtained using `lol_html_take_last_error` function.
+int lol_html_end_tag_after(
+    lol_html_end_tag_t *end_tag,
+    const char *content,
+    size_t content_len,
+    bool is_html
+);
+
+// Removes the end tag.
+void lol_html_end_tag_remove(lol_html_end_tag_t *end_tag);
+
+// Returns the end tag name.
+lol_html_str_t lol_html_end_tag_name_get(const lol_html_end_tag_t *end_tag);
+
+// Sets the tag name of the end tag.
+//
+// Name should be a valid UTF8-string.
+//
+// Returns 0 in case of success and -1 otherwise. The actual error message
+// can be obtained using `lol_html_take_last_error` function.
+int lol_html_end_tag_name_set(
+    lol_html_end_tag_t *end_tag,
+    const char *name,
+    size_t name_len
+);
+
 // Inserts the content at the end of the document, either as raw text or as HTML.
 //
 // The content should be a valid UTF-8 string.

c-api/src/element.rs

@@ -217,3 +217,64 @@ pub extern "C" fn lol_html_element_user_data_set(element: *mut Element, user_dat
 pub extern "C" fn lol_html_element_user_data_get(element: *mut Element) -> *mut c_void {
     get_user_data!(element)
 }
+
+type EndTagHandler = unsafe extern "C" fn(*mut EndTag, *mut c_void) -> RewriterDirective;
+
+#[no_mangle]
+pub extern "C" fn lol_html_element_on_end_tag(
+    element: *mut Element,
+    handler: EndTagHandler,
+    user_data: *mut c_void,
+) -> c_int {
+    let element = to_ref_mut!(element);
+    let () = unwrap_or_ret_err_code!(element.on_end_tag(move |end_tag| {
+        match unsafe { handler(end_tag, user_data) } {
+            RewriterDirective::Continue => Ok(()),
+            RewriterDirective::Stop => Err("The rewriter has been stopped.".into()),
+        }
+    }));
+    0
+}
+
+#[no_mangle]
+pub extern "C" fn lol_html_end_tag_before(
+    end_tag: *mut EndTag,
+    content: *const c_char,
+    content_len: size_t,
+    is_html: bool,
+) -> c_int {
+    content_insertion_fn_body! { end_tag.before(content, content_len, is_html) }
+}
+
+#[no_mangle]
+pub extern "C" fn lol_html_end_tag_after(
+    end_tag: *mut EndTag,
+    content: *const c_char,
+    content_len: size_t,
+    is_html: bool,
+) -> c_int {
+    content_insertion_fn_body! { end_tag.after(content, content_len, is_html) }
+}
+
+#[no_mangle]
+pub extern "C" fn lol_html_end_tag_remove(end_tag: *mut EndTag) {
+    to_ref_mut!(end_tag).remove();
+}
+
+#[no_mangle]
+pub extern "C" fn lol_html_end_tag_name_get(end_tag: *mut EndTag) -> Str {
+    let tag = to_ref_mut!(end_tag);
+    Str::new(tag.name())
+}
+
+#[no_mangle]
+pub extern "C" fn lol_html_end_tag_name_set(
+    end_tag: *mut EndTag,
+    name: *const c_char,
+    len: size_t,
+) -> c_int {
+    let tag = to_ref_mut!(end_tag);
+    let name = unwrap_or_ret_err_code! { to_str!(name, len) };
+    tag.set_name_str(name.to_string());
+    0
+}

c-api/tests/src/deps/picotest/picotest.c

@@ -26,6 +26,7 @@
 #include <stdio.h>
 #include <string.h>
 #include "picotest.h"
+#include "../../../../include/lol_html.h"
 
 struct test_t {
     int num_tests;
@@ -75,6 +76,16 @@ void _ok(int cond, const char *fmt, ...)
     fflush(stdout);
 }
 
+void _lol_ok(int cond, const char *file, int line) {
+    cond = !cond; // lol-html returns 0 on success
+    _ok(cond, "%s %d", file, line);
+    if (!cond) {
+        lol_html_str_t err = lol_html_take_last_error();
+        assert(err.data != NULL && err.len != 0);
+        printf("err: last lol_html err: %s", err.data);
+    }
+}
+
 int done_testing(void)
 {
     indent();

c-api/tests/src/deps/picotest/picotest.h

@@ -22,13 +22,17 @@
 #ifndef picotest_h
 #define picotest_h
 
+#include <assert.h>
+
 #ifdef __cplusplus
 extern "C" {
 #endif
 
 void note(const char *fmt, ...)  __attribute__((format (printf, 1, 2)));
 void _ok(int cond, const char *fmt, ...) __attribute__((format (printf, 2, 3)));
+void _lol_ok(int cond, const char *file, int line);
 #define ok(cond) _ok(cond, "%s %d", __FILE__, __LINE__)
+#define lol_ok(cond) _lol_ok(cond, __FILE__, __LINE__)
 int done_testing(void);
 void subtest(const char *name, void (*cb)(void));
 

c-api/tests/src/test_element_api.c

@@ -617,6 +617,48 @@ static void test_stop(lol_html_selector_t *selector, void *user_data) {
     ok(!err);
 }
 
+static lol_html_rewriter_directive_t modify_element_end_tag_name_inner(lol_html_end_tag_t *end_tag, void *user_data) {
+    int times_run = *(int*)user_data;
+
+    if (times_run == 0) {
+        lol_ok(lol_html_end_tag_before(end_tag, "!", 1, false));
+        const char *after_html = "<span>extra data</span>";
+        lol_ok(lol_html_end_tag_after(end_tag, after_html, strlen(after_html), true));
+
+        lol_html_str_t name = lol_html_end_tag_name_get(end_tag);
+        str_eq(name, "div");
+
+        lol_ok(lol_html_end_tag_name_set(end_tag, "div1", strlen("div1")));
+        name = lol_html_end_tag_name_get(end_tag);
+        str_eq(name, "div1");
+    } else {
+        lol_html_end_tag_remove(end_tag);
+    }
+
+    return LOL_HTML_CONTINUE;
+}
+
+static lol_html_rewriter_directive_t modify_element_end_tag_name_outer(
+    lol_html_element_t *element,
+    void *user_data
+) {
+    UNUSED(user_data);
+
+    static int times_run = -1; // so that it will be 0 on the first call to `inner`
+
+    lol_ok(lol_html_element_on_end_tag(element, modify_element_end_tag_name_inner, &times_run));
+    times_run += 1;
+
+    return LOL_HTML_CONTINUE;
+}
+
+EXPECT_OUTPUT(
+    modify_element_end_tag,
+    "<div>42!</div1><span>extra data</span><div>some data",
+    &EXPECTED_USER_DATA,
+    sizeof(EXPECTED_USER_DATA)
+);
+
 void element_api_test() {
     int user_data = 43;
 
@@ -718,4 +760,30 @@ void element_api_test() {
 
         lol_html_selector_free(selector);
     }
+
+    {
+        note("EndTagChange");
+
+        const char *selector_str = "div";
+        lol_html_selector_t *selector = lol_html_selector_parse(
+            selector_str,
+            strlen(selector_str)
+        );
+
+        lol_html_rewriter_builder_t *builder = lol_html_rewriter_builder_new();
+
+        lol_ok(lol_html_rewriter_builder_add_element_content_handlers(
+            builder,
+            selector,
+            modify_element_end_tag_name_outer,
+            NULL,
+            NULL,
+            NULL,
+            NULL,
+            NULL
+        ));
+
+        const char *input = "<div>42</div><div>some data</div>";
+        run_rewriter(builder, input, modify_element_end_tag, &user_data);
+    }
 }

src/base/bytes.rs

@@ -20,6 +20,15 @@ impl<'b> Bytes<'b> {
         encoding.encode(string).0.into()
     }
 
+    /// Same as `Bytes::from_str(&string).into_owned()`, but avoids copying in the common case where
+    /// the output and input encodings are the same.
+    pub fn from_string(string: String, encoding: &'static Encoding) -> Bytes<'static> {
+        Bytes(Cow::Owned(match encoding.encode(&string).0 {
+            Cow::Owned(bytes) => bytes,
+            Cow::Borrowed(_) => string.into_bytes(),
+        }))
+    }
+
     #[inline]
     pub fn from_str_without_replacements(
         string: &'b str,

src/rewritable_units/tokens/end_tag.rs

@@ -36,6 +36,11 @@ impl<'i> EndTag<'i> {
         self.raw = None;
     }
 
+    #[inline]
+    pub fn set_name_str(&mut self, name: String) {
+        self.set_name(Bytes::from_string(name, self.encoding))
+    }
+
     #[inline]
     pub fn before(&mut self, content: &str, content_type: ContentType) {
         self.mutations.before(content, content_type);