PawletOS/cromite
1
0
Fork 0
Code Issues Pull Requests Actions 3 Packages Projects Releases Wiki Activity
Files
032d2ef9f2e4f5c6bf204c5babf3d59cf5178cc4
cromite/patches/BRM047_Bromite-adblock-engine.patch
T

585 lines
18 KiB
Diff
Raw Blame History

From: csagan5 <32685696+csagan5@users.noreply.github.com>
Date: Wed, 28 Mar 2018 15:55:11 +0200
Subject: Bromite adblock engine
Ported from NoChromo patch
Make interception testable
Add domain support
Re-land: third-party filters support
---
net/BUILD.gn | 9 +-
net/url_request/adblock_intercept.cc | 324 ++++++++++++++++++++++++++++++++++
net/url_request/adblock_intercept.h | 16 ++
net/url_request/nochromo_intercept.cc | 116 ------------
net/url_request/nochromo_intercept.h | 13 --
net/url_request/url_request.cc | 19 +-
6 files changed, 364 insertions(+), 133 deletions(-)
create mode 100644 net/url_request/adblock_intercept.cc
create mode 100644 net/url_request/adblock_intercept.h
delete mode 100644 net/url_request/nochromo_intercept.cc
delete mode 100644 net/url_request/nochromo_intercept.h
diff --git a/net/BUILD.gn b/net/BUILD.gn
--- a/net/BUILD.gn
+++ b/net/BUILD.gn
@@ -1756,8 +1756,6 @@ component("net") {
"url_request/url_fetcher_response_writer.h",
"url_request/url_range_request_job.cc",
"url_request/url_range_request_job.h",
- "url_request/nochromo_intercept.cc",
- "url_request/nochromo_intercept.h",
"url_request/url_request.cc",
"url_request/url_request.h",
"url_request/url_request_context.cc",
@@ -1815,6 +1813,13 @@ component("net") {
"//third_party/zlib",
]
+ if (is_android) {
+ sources += [
+ "url_request/adblock_intercept.cc",
+ "url_request/adblock_intercept.h"
+ ]
+ }
+
if (enable_reporting) {
sources += [
"network_error_logging/network_error_logging_delegate.cc",
diff --git a/net/url_request/adblock_intercept.cc b/net/url_request/adblock_intercept.cc
new file mode 100644
--- /dev/null
+++ b/net/url_request/adblock_intercept.cc
@@ -0,0 +1,324 @@
+#include "url/gurl.h"
+
+#ifdef ADB_TESTER
+#include <cstddef>
+#include <cstdlib>
+#include <string.h>
+
+#include "log.h"
+#include <tld.h>
+
+#else
+
+#include "net/base/registry_controlled_domains/registry_controlled_domain.h"
+#include <android/log.h>
+
+#endif
+
+#include "net/url_request/adblock_entries.h"
+
+namespace net {
+
+const char *LOG_TAG = "Bromite";
+
+#ifdef ADB_TESTER
+int adblock_rules_count() { return ADBLOCK_ENTRY_COUNT; }
+#endif
+
+// True if the given canonical |host| is "[www.]<domain_in_lower_case>.<TLD>"
+// with a valid TLD. If |subdomain_permission| is ALLOW_SUBDOMAIN, we check
+// against host "*.<domain_in_lower_case>.<TLD>" instead. Will return the TLD
+// string in |tld|, if specified and the |host| can be parsed.
+static bool is_first_party(char *l_host, char *l_url_host) {
+ size_t tld_length;
+
+#ifdef ADB_TESTER
+ char *found_tld;
+
+ if (TLD_SUCCESS != tld_get_z(l_host, &found_tld))
+ return false;
+ tld_length = strlen(found_tld);
+ if (tld_length == 0)
+ return false;
+#else
+ tld_length = net::registry_controlled_domains::GetCanonicalHostRegistryLength(
+ l_host, net::registry_controlled_domains::EXCLUDE_UNKNOWN_REGISTRIES,
+ net::registry_controlled_domains::EXCLUDE_PRIVATE_REGISTRIES);
+ if ((tld_length == 0) || (tld_length == std::string::npos))
+ return false;
+#endif
+
+ int len = strlen(l_host);
+ char *tld = l_host + len - tld_length;
+
+ // Removes any subdomain from origin host.
+ int i = len - tld_length - 2, top_i = i;
+ if (i < 0) {
+ return false;
+ }
+ char *domain = l_host;
+ for (; i >= 0; i--) {
+ if (l_host[i] == '.') {
+ int p_len = top_i - i;
+ // skip "co" in "co.uk", "org" in "org.uk"
+ if (p_len <= 3) {
+ tld -= p_len + 1;
+ continue;
+ }
+
+ // segment is long enough, accept it at as a domain
+ domain = l_host + i;
+ len -= i;
+ break;
+ }
+ }
+
+#ifdef ADBLOCK_LOG
+ __android_log_print(ANDROID_LOG_INFO, LOG_TAG,
+ "%s: extracted domain suffix: \"%s\" (TLD=\"%s\")",
+ l_host, domain, tld);
+#endif
+
+ // Check if supplied URL host matches, including the dot.
+ int b_len = strlen(l_url_host);
+ if (b_len < len) {
+ return false;
+ }
+ for (int i = 0; i < len; i++) {
+ if (l_url_host[b_len - 1 - i] != domain[len - 1 - i])
+ return false;
+ }
+
+ // pass with flying colors
+ return true;
+}
+
+static char *strtolower(const char *str) {
+ int len = strlen(str);
+ char *ret = (char *)malloc(len + 1);
+ ret[len] = '\0';
+ for (int i = 0; i < len; i++) {
+ if ((65 <= str[i]) && (str[i] <= 90)) {
+ ret[i] = str[i] + 32;
+ } else {
+ ret[i] = str[i];
+ }
+ }
+ return ret;
+}
+
+static char *strtosep(const char *str) {
+ int len = strlen(str);
+ char *ret = (char *)malloc(len + 3);
+ ret[0] = '^';
+ ret[len + 1] = '^';
+ ret[len + 2] = '\0';
+ for (int i = 0; i < len; i++) {
+ if ((str[i] == ':') || (str[i] == '/') || (str[i] == '?') ||
+ (str[i] == '&') || (str[i] == '=')) {
+ ret[i + 1] = '^';
+ } else {
+ ret[i + 1] = str[i];
+ }
+ }
+ return ret;
+}
+
+static bool url_matches(const char *c_url, char *c_url_sep, char *c_url_lower,
+ char *c_url_lower_sep, adblock_entry *entry) {
+ bool match = false;
+ // select comparison string based on case and separator presence (separator
+ // takes some shortcuts)
+ bool match_case = ((entry->flags & ADBLOCK_FLAG_MATCH_CASE) != 0);
+ bool match_separator = ((entry->flags & ADBLOCK_FLAG_HAS_SEPARATOR) != 0);
+ const char *match_url =
+ match_case ? (match_separator ? c_url_sep : c_url)
+ : (match_separator ? c_url_lower_sep : c_url_lower);
+
+#ifdef ADBLOCK_LOG_MORE
+ __android_log_print(ANDROID_LOG_INFO, LOG_TAG, "[case:%d][sep:%d][%s]",
+ match_case, match_separator, match_url);
+#endif
+ // check for all match parts at >= position of last match
+ const char *last = match_url;
+ for (int m = 0; const char *url_match = entry->matches[m]; m++) {
+ bool is_last_match = entry->matches[m + 1] == NULL;
+ const char *pos = strstr(last, url_match);
+ match = (pos != NULL);
+
+#ifdef ADBLOCK_LOG_MORE
+ __android_log_print(ANDROID_LOG_INFO, LOG_TAG, "[%s][found:%d][match:%d]",
+ entry->matches[m], pos == NULL ? 0 : 1, match ? 1 : 0);
+#endif
+ // check if the url starts with the first match part
+ if (match && (m == 0) && ((entry->flags & ADBLOCK_FLAG_MATCH_BEGIN) != 0) &&
+ (pos != match_url))
+ match = false;
+
+ // check if the url ends with the last match part
+ if (match && is_last_match &&
+ ((entry->flags & ADBLOCK_FLAG_MATCH_END) != 0) &&
+ (pos != &match_url[strlen(match_url) - strlen(entry->matches[m])]))
+ match = false;
+
+ // check domain match
+ if (match && (m == 0) &&
+ ((entry->flags & ADBLOCK_FLAG_MATCH_DOMAIN) != 0) &&
+ (pos != match_url) && (pos[-1] != '^') && (pos[-1] != '.') &&
+ (pos[-1] != '/'))
+ match = false;
+
+ // short circuit
+ if (!match)
+ break;
+ }
+ return match;
+}
+
+bool url_match_domain(adblock_entry *entry, const std::string &origin_host) {
+ bool match_domain = true;
+ // check for a negative domain match
+ if (entry->domains_neg) {
+ if (origin_host.empty()) {
+ // skip this rule, cannot match on domain
+ return false;
+ }
+ for (int d = 0; const char *domain = entry->domains_neg[d]; d++) {
+ if (domain == origin_host) {
+ match_domain = false;
+ break;
+ }
+ }
+ }
+
+ // check for a required positive domain match
+ if (entry->domains) {
+ if (origin_host.empty()) {
+ // skip this rule, cannot match on domain
+ return false;
+ }
+ for (int d = 0; const char *domain = entry->domains[d]; d++) {
+ if (domain != origin_host) {
+ match_domain = false;
+ break;
+ }
+ }
+ }
+ return match_domain;
+}
+
+static bool url_match_party(adblock_entry *entry, const GURL &url,
+ const std::string &origin_host, bool &checked_fp,
+ bool &fp) {
+ bool wanted_fp;
+ if ((entry->flags & ADBLOCK_FLAG_THIRD_PARTY) != 0) {
+ wanted_fp = false;
+ } else if ((entry->flags & ADBLOCK_FLAG_FIRST_PARTY) != 0) {
+ wanted_fp = true;
+ } else {
+ // no-op
+ return true;
+ }
+
+ if (origin_host.empty()) {
+ // cannot match this rule, no origin host to determine first/third party
+ return false;
+ }
+
+#ifdef ADB_TESTER
+//__android_log_print(ANDROID_LOG_INFO, LOG_TAG, "matchFirst=%d matchThird=%d",
+// matchFirstParty, matchThirdParty);
+#endif
+ if (!checked_fp) {
+ // lower-case version
+ char *l_host = strtolower(origin_host.c_str()),
+ *l_url_host = strtolower(url.host().c_str());
+
+ // is the URL a first-party to the current page's host?
+ fp = is_first_party(l_host, l_url_host);
+
+ checked_fp = true;
+#ifdef ADB_TESTER
+ __android_log_print(ANDROID_LOG_INFO, LOG_TAG,
+ "is_first_party(\"%s\", \"%s\") = %s", l_host,
+ l_url_host, fp ? "true" : "false");
+#endif
+ free(l_host);
+ free(l_url_host);
+ }
+
+ return fp == wanted_fp;
+}
+
+int adblock_intercept(const GURL &url, const std::string &origin_host) {
+ // if (!url.is_valid() || !url.SchemeIsHTTPOrHTTPS()) {
+ // return 0;
+ // }
+
+ if (url.is_valid() && url.SchemeIsHTTPOrHTTPS()) {
+ const char *c_url = url.spec().c_str();
+ char *c_url_lower = strtolower(c_url);
+ char *c_url_sep = strtosep(c_url);
+ char *c_url_lower_sep = strtosep(c_url_lower);
+
+#ifdef ADBLOCK_LOG
+ __android_log_print(ANDROID_LOG_INFO, LOG_TAG, "[%s with host '%s'] [%s]",
+ c_url, url.host().c_str(), origin_host.c_str());
+#endif
+
+ bool checked_fp = false, fp = false;
+
+ bool intercept = false;
+ for (int i = 0; i < ADBLOCK_ENTRY_COUNT; i++) {
+ adblock_entry *entry = &ADBLOCK_ENTRIES[i];
+
+ // no use checking rules when we're intercepting, or exceptions when not
+ bool check =
+ (!intercept && ((entry->flags & ADBLOCK_FLAG_EXCEPTION) == 0)) ||
+ (intercept && ((entry->flags & ADBLOCK_FLAG_EXCEPTION) != 0));
+ if (!check)
+ continue;
+
+ // first check for domain matches, a quick branch out if matching
+ if (!url_match_domain(entry, origin_host))
+ continue;
+
+ // check on the URL matcher
+ if (!url_matches(c_url, c_url_sep, c_url_lower, c_url_lower_sep, entry))
+ continue;
+
+ // finally check first/third-party
+ if (!url_match_party(entry, url, origin_host, checked_fp, fp))
+ continue;
+
+#ifdef ADBLOCK_LOG
+ if (!intercept) {
+ __android_log_print(ANDROID_LOG_INFO, LOG_TAG,
+ "--> intercept (#%d: \"%s\") (%x)", i,
+ entry->matches[0], entry->flags);
+ } else {
+ __android_log_print(ANDROID_LOG_INFO, LOG_TAG, "--> pass (%d) (#%d)", i,
+ entry->flags);
+ }
+#endif
+ intercept = !intercept;
+ } // for each entry
+
+ free(c_url_sep);
+ free(c_url_lower);
+ free(c_url_lower_sep);
+
+ if (intercept) {
+#ifdef ADBLOCK_LOG
+ __android_log_print(ANDROID_LOG_INFO, LOG_TAG, "blocked");
+#endif
+ return 1;
+ }
+#ifdef ADBLOCK_LOG
+ __android_log_print(ANDROID_LOG_INFO, LOG_TAG, "pass");
+#endif
+ }
+ return 0;
+}
+
+} // namespace net
diff --git a/net/url_request/adblock_intercept.h b/net/url_request/adblock_intercept.h
new file mode 100644
--- /dev/null
+++ b/net/url_request/adblock_intercept.h
@@ -0,0 +1,16 @@
+#ifndef NET_URL_REQUEST_ADBLOCK_INTERCEPT_H_
+#define NET_URL_REQUEST_ADBLOCK_INTERCEPT_H_
+
+#include "url/gurl.h"
+
+namespace net {
+
+#ifdef ADB_TESTER
+int adblock_rules_count();
+#endif
+
+int adblock_intercept(const GURL &url, const std::string &origin_host);
+
+} // namespace net
+
+#endif // NET_URL_REQUEST_ADBLOCK_INTERCEPT_H_
diff --git a/net/url_request/nochromo_intercept.cc b/net/url_request/nochromo_intercept.cc
deleted file mode 100644
--- a/net/url_request/nochromo_intercept.cc
+++ /dev/null
@@ -1,116 +0,0 @@
-#include "url/gurl.h"
-#include "net/url_request/nochromo_entries.h"
-
-#include <android/log.h>
-
-namespace net {
-
-#define NOCHROMO_LOG 0
-#define NOCHROMO_LOG_MORE 0
-
-static char* strtolower(const char* str) {
- int len = strlen(str);
- char* ret = (char*)malloc(len + 1);
- ret[len] = '\0';
- for (int i = 0; i < len; i++) {
- if ((65 <= str[i]) && (str[i] <= 90)) {
- ret[i] = str[i] + 32;
- } else {
- ret[i] = str[i];
- }
- }
- return ret;
-}
-
-static char* strtosep(const char* str) {
- int len = strlen(str);
- char* ret = (char*)malloc(len + 3);
- ret[0] = '^';
- ret[len + 1] = '^';
- ret[len + 2] = '\0';
- for (int i = 0; i < len; i++) {
- if ((str[i] == ':') || (str[i] == '/') || (str[i] == '?') || (str[i] == '&') || (str[i] == '=')) {
- ret[i + 1] = '^';
- } else {
- ret[i + 1] = str[i];
- }
- }
- return ret;
-}
-
-int nochromo_intercept(const GURL& url) {
- if (url.is_valid() && url.SchemeIsHTTPOrHTTPS()) {
- const char* c_url = url.spec().c_str();
- char* c_url_lower = strtolower(c_url);
- char* c_url_sep = strtosep(c_url);
- char* c_url_lower_sep = strtosep(c_url_lower);
-
- if (NOCHROMO_LOG) __android_log_print(ANDROID_LOG_INFO, "NoChromo", "[%s]", c_url);
-
- bool intercept = false;
- for (int i = 0; i < NOCHROMO_ENTRY_COUNT; i++) {
- nochromo_entry* entry = &NOCHROMO_ENTRIES[i];
-
- // no use checking rules when we're intercepting, or exceptions when not
- bool check =
- (!intercept && ((entry->flags & NOCHROMO_FLAG_EXCEPTION) == 0)) ||
- (intercept && ((entry->flags & NOCHROMO_FLAG_EXCEPTION) != 0));
-
- if (check) {
- bool match = false;
-
- // select comparison string based on case and separator presence (separator takes some shortcuts)
- bool match_case = ((entry->flags & NOCHROMO_FLAG_MATCH_CASE) != 0);
- bool match_separator = ((entry->flags & NOCHROMO_FLAG_HAS_SEPARATOR) != 0);
- const char* match_url = match_case ? (match_separator ? c_url_sep : c_url) : (match_separator ? c_url_lower_sep : c_url_lower);
-
- if (NOCHROMO_LOG || NOCHROMO_LOG_MORE) __android_log_print(ANDROID_LOG_INFO, "NoChromo", "[case:%d][sep:%d][%s]", match_case, match_separator, match_url);
-
- // check for all match parts at >= position of last match
- const char* last = match_url;
- for (int m = 0; m < entry->matchcount; m++) {
- const char* pos = strstr(last, entry->matches[m]);
- match = (pos != NULL);
-
- if (NOCHROMO_LOG || NOCHROMO_LOG_MORE) __android_log_print(ANDROID_LOG_INFO, "NoChromo", "[%s][found:%d][match:%d]", entry->matches[m], pos == NULL ? 0 : 1, match ? 1 : 0);
-
- // check if the url starts with the first match part
- if (match && (m == 0) && ((entry->flags & NOCHROMO_FLAG_MATCH_BEGIN) != 0) && (pos != match_url)) match = false;
-
- // check if the url ends with the last match part
- if (match && (m == entry->matchcount - 1) && ((entry->flags & NOCHROMO_FLAG_MATCH_END) != 0) && (pos != &match_url[strlen(match_url) - strlen(entry->matches[m])])) match = false;
-
- // check domain match
- if (match && (m == 0) && ((entry->flags & NOCHROMO_FLAG_MATCH_DOMAIN) != 0) && (pos != match_url) && (pos[-1] != '^') && (pos[-1] != '.') && (pos[-1] != '/')) match = false;
-
- // short circuit
- if (!match) break;
- }
-
- if (match) {
- if (NOCHROMO_LOG) {
- if (!intercept) {
- __android_log_print(ANDROID_LOG_INFO, "NoChromo", "--> intercept (%d) (%d)", i, entry->flags);
- } else {
- __android_log_print(ANDROID_LOG_INFO, "NoChromo", "--> pass (%d) (%d)", i, entry->flags);
- }
- }
- intercept = !intercept;
- }
- }
- }
-
- free(c_url_sep);
- free(c_url_lower);
- free(c_url_lower_sep);
-
- if (intercept) {
- if (NOCHROMO_LOG) __android_log_print(ANDROID_LOG_INFO, "NoChromo", "intercepted!");
- return 1;
- }
- if (NOCHROMO_LOG) __android_log_print(ANDROID_LOG_INFO, "NoChromo", "pass!");
- }
- return 0;
-}
-
-} // namespace net
\ No newline at end of file
diff --git a/net/url_request/nochromo_intercept.h b/net/url_request/nochromo_intercept.h
deleted file mode 100644
--- a/net/url_request/nochromo_intercept.h
+++ /dev/null
@@ -1,13 +0,0 @@
-#ifndef NET_URL_REQUEST_NOCHROMO_INTERCEPT_H_
-#define NET_URL_REQUEST_NOCHROMO_INTERCEPT_H_
-
-#include "url/gurl.h"
-
-namespace net {
-
-GURL nochromo_intercepted = GURL("http://127.0.0.1/");
-int nochromo_intercept(const GURL& url);
-
-} // namespace net
-
-#endif // NET_URL_REQUEST_NOCHROMO_INTERCEPT_H_
diff --git a/net/url_request/url_request.cc b/net/url_request/url_request.cc
--- a/net/url_request/url_request.cc
+++ b/net/url_request/url_request.cc
@@ -44,7 +44,9 @@
#include "url/gurl.h"
#include "url/origin.h"
-#include "net/url_request/nochromo_intercept.h"
+#if defined(OS_ANDROID)
+#include "net/url_request/adblock_intercept.h"
+#endif
#if BUILDFLAG(ENABLE_REPORTING)
#include "net/network_error_logging/network_error_logging_service.h"
@@ -559,7 +561,9 @@ URLRequest::URLRequest(const GURL& url,
: context->network_delegate()),
net_log_(NetLogWithSource::Make(context->net_log(),
NetLogSourceType::URL_REQUEST)),
+#if !defined(OS_ANDROID)
url_chain_(1, url),
+#endif
attach_same_site_cookies_(false),
method_("GET"),
referrer_policy_(CLEAR_REFERRER_ON_TRANSITION_FROM_SECURE_TO_INSECURE),
@@ -585,7 +589,18 @@ URLRequest::URLRequest(const GURL& url,
// Sanity check out environment.
DCHECK(base::ThreadTaskRunnerHandle::IsSet());
- if (nochromo_intercept(url)) url_chain_.push_back(nochromo_intercepted);
+#if defined(OS_ANDROID)
+ std::string initiatorHost;
+ if (initiator_.has_value()) {
+ initiatorHost = initiator_.value().host();
+ }
+
+ if (adblock_intercept(url, initiatorHost)) {
+ url_chain_ = { url, GURL("http://127.0.0.1") };
+ } else {
+ url_chain_ = { url };
+ }
+#endif
context->url_requests()->insert(this);
net_log_.BeginEvent(
--
2.7.4
Reference in New Issue View Git Blame Copy Permalink