fastemailparser 0.1.0__tar.gz → 0.1.2__tar.gz

fastemailparser-0.1.2/MANIFEST.in

@@ -0,0 +1,2 @@
+include email.h
+recursive-include src *.h

{fastemailparser-0.1.0/fastemailparser.egg-info → fastemailparser-0.1.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: fastemailparser
-Version: 0.1.0
+Version: 0.1.2
 Summary: Very fast email parsing tool, split emails, retrieve headers & signatures
 Home-page: https://github.com/Methode-dev/EmailParser
 Author: Julien Calenge @ Méthode

fastemailparser-0.1.2/email.h

@@ -0,0 +1,12 @@
+#ifndef EMAIL
+#define EMAIL
+
+typedef struct email_s {
+    int last_index;
+    int exhausted;
+    int yield_if_empty_chain; /* yield body even if no separator found (single
+                                 email) */
+    char *body;
+} email_t;
+
+#endif

{fastemailparser-0.1.0 → fastemailparser-0.1.2/fastemailparser.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: fastemailparser
-Version: 0.1.0
+Version: 0.1.2
 Summary: Very fast email parsing tool, split emails, retrieve headers & signatures
 Home-page: https://github.com/Methode-dev/EmailParser
 Author: Julien Calenge @ Méthode

{fastemailparser-0.1.0 → fastemailparser-0.1.2}/fastemailparser.egg-info/SOURCES.txt

@@ -1,5 +1,7 @@
 LICENSE
+MANIFEST.in
 README.md
+email.h
 emailparser.c
 setup.py
 fastemailparser.egg-info/PKG-INFO
@@ -7,9 +9,17 @@ fastemailparser.egg-info/SOURCES.txt
 fastemailparser.egg-info/dependency_links.txt
 fastemailparser.egg-info/top_level.txt
 src/body.c
+src/body.h
+src/buf.h
 src/email_iter.c
+src/email_iter.h
 src/headers.c
+src/headers.h
 src/html.c
+src/html.h
 src/mime.c
+src/mime.h
 src/signature.c
-src/standalone.c
+src/signature.h
+src/standalone.c
+src/standalone.h

{fastemailparser-0.1.0 → fastemailparser-0.1.2}/setup.py

@@ -28,7 +28,7 @@ module = Extension(
 
 setup(
     name="fastemailparser",
-    version="0.1.0",
+    version="0.1.2",
     author="Julien Calenge @ Méthode",
     author_email="julien.calenge@methode.dev",
     description="Very fast email parsing tool, split emails, retrieve headers & signatures",

fastemailparser-0.1.2/src/body.h

@@ -0,0 +1,9 @@
+#ifndef EMAILPARSER_BODY_H
+#define EMAILPARSER_BODY_H
+#define PY_SSIZE_T_CLEAN
+#include <Python.h>
+#include <stddef.h>
+
+const char *find_body_start(const char *text, size_t len);
+PyObject *py_extract_body(PyObject *module, PyObject *args);
+#endif

fastemailparser-0.1.2/src/buf.h

@@ -0,0 +1,25 @@
+#ifndef EMAILPARSER_BUF_H
+#define EMAILPARSER_BUF_H
+#include <stdlib.h>
+#include <string.h>
+
+typedef struct {
+    char *buf;
+    size_t len, cap;
+} strbuf_t;
+
+static inline int sb_push(strbuf_t *sb, const char *s, size_t n) {
+    if (sb->len + n + 1 > sb->cap) {
+        size_t cap = (sb->len + n + 1) * 2;
+        char *tmp = realloc(sb->buf, cap);
+        if (!tmp)
+            return -1;
+        sb->buf = tmp;
+        sb->cap = cap;
+    }
+    memcpy(sb->buf + sb->len, s, n);
+    sb->len += n;
+    sb->buf[sb->len] = '\0';
+    return 0;
+}
+#endif

fastemailparser-0.1.2/src/email_iter.h

@@ -0,0 +1,16 @@
+#ifndef EMAILPARSER_EMAIL_ITER_H
+#define EMAILPARSER_EMAIL_ITER_H
+#include "email.h"
+#include <stdbool.h>
+
+#ifndef SEPARATOR_REGEX
+#define SEPARATOR_REGEX_GEN_EN "(From|Sent|To|Subject|Cc|Bcc) ?(&nbsp;:|:) ?"
+#define SEPARATOR_REGEX_GEN_FR "(De|À|Envoyé|Objet|Cc|Cci) ?(&nbsp;:|:) ?"
+#define SEPARATOR_REGEX_STA_ALL "(De|From) ?(&nbsp;:|:) ?"
+#define SEPARATOR_REGEX_END_ALL "(Objet|Subject) ?(&nbsp;:|:) ?"
+#define SEPARATOR_REGEX SEPARATOR_REGEX_STA_ALL
+#endif
+
+email_t *new_email(char *raw);
+bool get_next_val(email_t *email);
+#endif

fastemailparser-0.1.2/src/headers.h

@@ -0,0 +1,9 @@
+#ifndef EMAILPARSER_HEADERS_H
+#define EMAILPARSER_HEADERS_H
+#define PY_SSIZE_T_CLEAN
+#include <Python.h>
+#include <stddef.h>
+
+const char *canonical_key(const char *name, size_t len);
+PyObject *py_parse_headers(PyObject *module, PyObject *args);
+#endif

fastemailparser-0.1.2/src/html.h

@@ -0,0 +1,30 @@
+#ifndef EMAILPARSER_HTML_H
+#define EMAILPARSER_HTML_H
+#define PY_SSIZE_T_CLEAN
+#include <Python.h>
+#include "buf.h"
+#include <ctype.h>
+#include <libxml/tree.h>
+
+int walk_text(xmlNodePtr node, strbuf_t *sb);
+PyObject *segment_to_text(const char *html);
+char *html_to_plain_c(const char *html);
+
+/* Detect HTML: <tag>, </tag> — not bare < in email addresses */
+static inline int looks_like_html(const char *text, size_t len) {
+    for (size_t i = 0; i + 1 < len && i < 512; i++) {
+        if (text[i] != '<')
+            continue;
+        if (text[i + 1] == '/')
+            return 1;
+        if (isalpha((unsigned char)text[i + 1])) {
+            size_t j = i + 2;
+            while (j < len && isalpha((unsigned char)text[j]))
+                j++;
+            if (j < len && (text[j] == '>' || text[j] == ' ' || text[j] == '/'))
+                return 1;
+        }
+    }
+    return 0;
+}
+#endif

fastemailparser-0.1.2/src/mime.h

@@ -0,0 +1,8 @@
+#ifndef EMAILPARSER_MIME_H
+#define EMAILPARSER_MIME_H
+#include <stddef.h>
+
+char *decode_qp(const char *in, size_t in_len, size_t *out_len);
+char *skip_mime_headers(char *raw);
+int has_html_mime_part(const char *text, int len);
+#endif

fastemailparser-0.1.2/src/signature.h

@@ -0,0 +1,8 @@
+#ifndef EMAILPARSER_SIGNATURE_H
+#define EMAILPARSER_SIGNATURE_H
+#define PY_SSIZE_T_CLEAN
+#include <Python.h>
+
+PyObject *py_find_signature(PyObject *module, PyObject *args);
+PyObject *py_strip_signature(PyObject *module, PyObject *args);
+#endif

fastemailparser-0.1.2/src/standalone.h

@@ -0,0 +1,10 @@
+#ifndef EMAILPARSER_STANDALONE_H
+#define EMAILPARSER_STANDALONE_H
+#define PY_SSIZE_T_CLEAN
+#include <Python.h>
+#include <stddef.h>
+
+char *extract_css(const char *html, size_t html_len);
+PyObject *wrap_standalone(const char *css, size_t css_len, const char *segment,
+                          size_t seg_len);
+#endif