KAZ/depollueur
7
1
Fork 0
Code Issues Pull Requests Projects Releases Activity

Compare commits

base: KAZ:99e888f5d115512cb8c1ecde09a2b6718f74ba86
Branches Tags
KAZ:master KAZ:test-skip KAZ:tests KAZ:develop
...
compare: KAZ:develop
Branches Tags
KAZ:master KAZ:test-skip KAZ:tests KAZ:develop

4 Commits
99e888f5d1 ... develop

Author SHA1 Message Date
François
29480b299c fix boundary 2022-11-20 14:55:44 +01:00
François
6def55c5a7 typo 2022-10-31 19:15:51 +01:00
François
c448478c39 add lang in a.php 2022-10-31 07:40:19 +01:00
François
b4ea2185f0 fix synchro plain and html 2022-10-31 07:08:19 +01:00
9 changed files with 199 additions and 104 deletions
Expand all files Collapse all files

11
README.md
View File

@ -41,3 +41,14 @@ git clone https://git.kaz.bzh/KAZ/depollueur.git
make -j $(nproc) make -j $(nproc)
``` ```
##Astuces
Pour la mise au point dans Kaz, il faut bloquer la mise à jour automatique avec :
``` bash
touch /kaz/git/depollueur/dontupdage
/kaz/bin/container.sh stop jirafeau postfix; docker system prune; /kaz/bin/install.sh jirafeau postfix
```

46
src/Jirafeau/a.php
View File

@ -12,6 +12,7 @@
a.php?s=mel@domain.org&t=password + [action] => manage account a.php?s=mel@domain.org&t=password + [action] => manage account
action: a=login a=logout a=r[on|off] a=p[minute|hour|day|week|month|quarter] action: a=login a=logout a=r[on|off] a=p[minute|hour|day|week|month|quarter]
*/ */
use PHPMailer\PHPMailer\PHPMailer; use PHPMailer\PHPMailer\PHPMailer;
use PHPMailer\PHPMailer\SMTP; use PHPMailer\PHPMailer\SMTP;
use PHPMailer\PHPMailer\Exception; use PHPMailer\PHPMailer\Exception;
@ -45,7 +46,7 @@ define ('E_BAD_ARCHIVE_NAME', 'Bad archive name format');
define ('E_CREATE_ZIP', "Impossible de créer l'archive."); define ('E_CREATE_ZIP', "Impossible de créer l'archive.");
define ('E_OPEN_ZIP', "Impossible d'ouvrir l'archive."); define ('E_OPEN_ZIP', "Impossible d'ouvrir l'archive.");
define ('M_BAD_KEY', "Mauvaise clef pour "); define ('M_BAD_KEY', "Mauvaise clef pour ");
define ('M_BAD_SENDER_NAME', 'Votre mèl est incorrect'); define ('M_BAD_SENDER_NAME', 'Votre mél est incorrect');
define ('M_BAD_TOKEN', "Vous n'utilisez pas le bon jeton (consultez votre messagerie)."); define ('M_BAD_TOKEN', "Vous n'utilisez pas le bon jeton (consultez votre messagerie).");
define ('M_TOO_LONG_BEFORE_LOGGED', "Jeton de connexion trop ancien."); define ('M_TOO_LONG_BEFORE_LOGGED', "Jeton de connexion trop ancien.");
define ('M_TOO_LONG_LOGGED', "Temps de connexion dépassé."); define ('M_TOO_LONG_LOGGED', "Temps de connexion dépassé.");
@ -60,7 +61,7 @@ define ('M_FILES_RENAMED', " fichiers renommés.");
define ('M_FILE_NOT_FOUND', " fichier est expiré."); define ('M_FILE_NOT_FOUND', " fichier est expiré.");
define ('M_FILE_RENAMED', " fichier renommé."); define ('M_FILE_RENAMED', " fichier renommé.");
define ('M_INTRO_FORM', "Quelles sont les informations me concernant ?"); define ('M_INTRO_FORM', "Quelles sont les informations me concernant ?");
define ('M_MEL', "votre mèl"); define ('M_MEL', "votre mél");
define ('M_NO_FILENAME', 'SansNom'); define ('M_NO_FILENAME', 'SansNom');
define ('M_NO_SENDER', 'kaz-'); define ('M_NO_SENDER', 'kaz-');
define ('M_NEW_ATTACHEMENT_DIRNAME', "nouveau"); define ('M_NEW_ATTACHEMENT_DIRNAME', "nouveau");
@ -155,6 +156,19 @@ if (isset ($_REQUEST [A_PERIOD]) && !empty ($_REQUEST [A_PERIOD])) {
exit; exit;
} }
// ========================================
if (isset ($_REQUEST [A_LANG]) && !empty ($_REQUEST [A_LANG])) {
if (!preg_match ("/^([a-z0-9\+_\-]+)(\.[a-z0-9\+_\-]+)*@([a-z0-9\-]+\.)+[a-z]{2,6}$/i", $_REQUEST [A_LANG]))
$content = DEFAULT_LANG.NL;
else
$content = getSenderLang ($_REQUEST [A_LANG]).NL;
header ('HTTP/1.0 200 OK');
header ('Content-Length: ' . strlen ($content));
header ('Content-Type: text/plain');
echo $content;
exit;
}
// ======================================== // ========================================
$doUpdate = false; $doUpdate = false;
if (isset ($_REQUEST [A_UPDATE]) && !empty ($_REQUEST [A_UPDATE])) { if (isset ($_REQUEST [A_UPDATE]) && !empty ($_REQUEST [A_UPDATE])) {
@ -748,10 +762,17 @@ if ($doDownload) {
$modif = true; $modif = true;
continue; continue;
} }
$srcName = $dstName = ($link ['file_name'] ? $link ['file_name'] : M_NO_FILENAME); $dstName = ($link ['file_name'] ? $link ['file_name'] : M_NO_FILENAME);
if (in_array ($srcName, $singleName)) if (in_array ($dstName, $singleName)) {
for ($i = 0; $i < 10000; ++$i) { $dstFilename = $dstName;
$dstName = sprintf ("%s-%2d", $srcName, $i); $dstExtension = "";
$dstExtensionPos = strrpos ($dstName, '.');
if ($dstExtensionPos) {
$dstFilename = substr ($dstName, 0, $dstExtensionPos);
$dstExtension = substr ($dstName, $dstExtensionPos);
}
for ($i = 1; $i < 10000; ++$i) {
$dstName = sprintf ("%s-%02d%s", $dstFilename, $i, $dstExtension);
if (!in_array ($dstName, $singleName)) { if (!in_array ($dstName, $singleName)) {
if (isset ($archiveContent [T_RENAME])) if (isset ($archiveContent [T_RENAME]))
++$archiveContent [T_RENAME]; ++$archiveContent [T_RENAME];
@ -761,6 +782,7 @@ if ($doDownload) {
break; break;
} }
} }
}
$singleName [] = $dstName; $singleName [] = $dstName;
$archiveContent [$cat][T_ENTRIES][] = [T_HASH => $link ['hash'], T_FILENAME => $dstName, T_CRYPT_KEY => $cryptKey, T_CRYPTED => $link ['crypted']]; $archiveContent [$cat][T_ENTRIES][] = [T_HASH => $link ['hash'], T_FILENAME => $dstName, T_CRYPT_KEY => $cryptKey, T_CRYPTED => $link ['crypted']];
} }
@ -871,7 +893,7 @@ if (isset ($_REQUEST [A_ACTION]) && $_REQUEST [A_ACTION] == T_LOGIN && $sender)
echo M_SEND_TOKEN; echo M_SEND_TOKEN;
else else
echo echo
"Erreur dans l'envoi. V&eacute;ritiez votre m&egrave;l."; "Erreur dans l'envoi. V&eacute;ritiez votre m&eacute;l.";
} }
echo "<br/><br/><br/>"; echo "<br/><br/><br/>";
require (JIRAFEAU_ROOT.'lib/template/footer.php'); require (JIRAFEAU_ROOT.'lib/template/footer.php');
@ -1065,6 +1087,7 @@ div.frame {border: 1px; border-style: solid; padding: 1em; margin: 1em;}
$defaultChecked = []; $defaultChecked = [];
$defaultChecked [isSenderTrack ($sender) ? "on" : "off"] = ' checked="checked"'; $defaultChecked [isSenderTrack ($sender) ? "on" : "off"] = ' checked="checked"';
$defaultChecked [getSenderPeriod ($sender)] = ' selected="selected"'; $defaultChecked [getSenderPeriod ($sender)] = ' selected="selected"';
$defaultChecked [getSenderLang ($sender)] = ' selected="selected"';
echo echo
'<form method="post">'. '<form method="post">'.
'Je veux que Kaz suive tous mes futurs envois: '. 'Je veux que Kaz suive tous mes futurs envois: '.
@ -1080,6 +1103,15 @@ echo
'<select name="'.A_ACTION.'" style="width: auto !important;">'; '<select name="'.A_ACTION.'" style="width: auto !important;">';
foreach ($periodText as $item => $text) foreach ($periodText as $item => $text)
echo ' <option value="'.A_PERIOD.$item.'"'.$defaultChecked [$item].'>'.$text.'</option>'; echo ' <option value="'.A_PERIOD.$item.'"'.$defaultChecked [$item].'>'.$text.'</option>';
echo
'</select> '.
'<button type="submit">'."valider".'</button>'.
'</form>'.
'<form method="post">'.
'Je préfère envoyer mes messages en langue '.
'<select name="'.A_ACTION.'" style="width: auto !important;">';
foreach ($langText as $item => $text)
echo ' <option value="'.A_LANG.$item.'"'.$defaultChecked [$item].'>'.$text.'</option>';
echo echo
'</select> '. '</select> '.
'<button type="submit">'."valider".'</button>'. '<button type="submit">'."valider".'</button>'.

20
src/bash/filter.sh
View File

@ -66,6 +66,7 @@ ARCHIVE_TITLE="archive_content"
ARCHIVE_MIME="text/kaz_email_archive" ARCHIVE_MIME="text/kaz_email_archive"
KEEP_FAILED=true KEEP_FAILED=true
DEBUG=
#################### FONCTIONS ############################################ #################### FONCTIONS ############################################
BOLD='' BOLD=''
@ -134,7 +135,7 @@ ARCHIVE_CONTENT="${REP_PIECE_JOINTE}/archive-content.txt"
JIRAFEAU_ERROR="${REP_PIECE_JOINTE}/jirafeau-error.txt" JIRAFEAU_ERROR="${REP_PIECE_JOINTE}/jirafeau-error.txt"
# Clean up when done or when aborting. # Clean up when done or when aborting.
trap "rm -rf in.$$ in.$$.altered ${REP_PIECE_JOINTE}" 0 1 2 3 15 [ -z "${DEBUG}" ] && trap "rm -rf in.$$ in.$$.altered ${REP_PIECE_JOINTE}" 0 1 2 3 15
if ! cat > "in.$$"; then if ! cat > "in.$$"; then
LOG_FIC "${RED}Cannot save mail to file${NC}" LOG_FIC "${RED}Cannot save mail to file${NC}"
@ -142,8 +143,7 @@ if ! cat > "in.$$"; then
fi fi
LOG_FIC "\n" \ LOG_FIC "\n" \
" size: ${YELLOW}$(wc -c < "${INSPECT_DIR}/in.$$")${NC}" " size: ${YELLOW}$(wc -c < "${INSPECT_DIR}/in.$$")${NC}"
# XXX trace [ -n "${DEBUG}" ] && (mkdir -p "${DIR_LOG}/pb/" ; cp "${INSPECT_DIR}/in.$$" "${DIR_LOG}/pb/in.$$.orig")
# cp "${INSPECT_DIR}/in.$$" "${INSPECT_DIR}/in.$$.bak"
mkdir -p "${REP_PIECE_JOINTE}/" mkdir -p "${REP_PIECE_JOINTE}/"
>"${OLD_LINKS}" >"${OLD_LINKS}"
@ -156,8 +156,7 @@ echo "time: ${DATE_TEMPS}\nid: $(date +%s)" > "${ARCHIVE_CONTENT}"
LOG_FIC "${CYAN}${SHRINK_CMD} -u \"${INSPECT_DIR}/in.$$\" 2>> \"${FIC_LOG}\" > \"${OLD_LINKS}\"${NC}" LOG_FIC "${CYAN}${SHRINK_CMD} -u \"${INSPECT_DIR}/in.$$\" 2>> \"${FIC_LOG}\" > \"${OLD_LINKS}\"${NC}"
"${SHRINK_CMD}" -u "${INSPECT_DIR}/in.$$" 2>> "${FIC_LOG}" > "${OLD_LINKS}" "${SHRINK_CMD}" -u "${INSPECT_DIR}/in.$$" 2>> "${FIC_LOG}" > "${OLD_LINKS}"
cat "${OLD_LINKS}" | grep "${JIRAFEAU_URL}" | while read REMOTE_LINK cat "${OLD_LINKS}" | grep "${JIRAFEAU_URL}" | while read REMOTE_LINK; do
do
REMOTE_REF=$(echo "${REMOTE_LINK}" | sed -e 's/.*h=\([^&]*\).*/\1/' -e 's/.*http.*//') REMOTE_REF=$(echo "${REMOTE_LINK}" | sed -e 's/.*h=\([^&]*\).*/\1/' -e 's/.*http.*//')
[ -z "${REMOTE_REF}" ] && continue [ -z "${REMOTE_REF}" ] && continue
REMOTE_KEY=$(echo "${REMOTE_LINK}" | grep "k=" | sed 's%.*k=\([^&]*\).*%\1%') REMOTE_KEY=$(echo "${REMOTE_LINK}" | grep "k=" | sed 's%.*k=\([^&]*\).*%\1%')
@ -171,10 +170,8 @@ LOG_FIC " - archive starts with: ${NL}${YELLOW}$(cat ${ARCHIVE_CONTENT})${NC}"
# Etape extraction des pieces jointes # Etape extraction des pieces jointes
LOG_FIC "${CYAN}${SHRINK_CMD} -s ${MAX_KEEP_IN_MAIL} -d ${REP_PIECE_JOINTE} ${INSPECT_DIR}/in.$$ ${NC}" LOG_FIC "${CYAN}${SHRINK_CMD} -s ${MAX_KEEP_IN_MAIL} -d ${REP_PIECE_JOINTE} ${INSPECT_DIR}/in.$$ ${NC}"
"${SHRINK_CMD}" -s "${MAX_KEEP_IN_MAIL}" -d "${REP_PIECE_JOINTE}" "${INSPECT_DIR}/in.$$" 2>> "${FIC_LOG}" | { "${SHRINK_CMD}" -s "${MAX_KEEP_IN_MAIL}" -d "${REP_PIECE_JOINTE}" "${INSPECT_DIR}/in.$$" 2>> "${FIC_LOG}" | {
while read ATTACH_TMP_NAME while read ATTACH_TMP_NAME; do
do if [ -d "${ATTACH_TMP_NAME}" ]; then
if [ -d "${ATTACH_TMP_NAME}" ]
then
ATTACH_MEDIA="${ATTACH_TMP_NAME}/media" ATTACH_MEDIA="${ATTACH_TMP_NAME}/media"
ATTACH_NAME=$(grep "^Name: " "${ATTACH_TMP_NAME}/meta" | cut -c 7- ) ATTACH_NAME=$(grep "^Name: " "${ATTACH_TMP_NAME}/meta" | cut -c 7- )
ATTACH_CONTENT_TYPE=$(grep "^Content-Type: " "${ATTACH_TMP_NAME}/meta" | cut -c 15- ) ATTACH_CONTENT_TYPE=$(grep "^Content-Type: " "${ATTACH_TMP_NAME}/meta" | cut -c 15- )
@ -195,7 +192,7 @@ LOG_FIC "${CYAN}${SHRINK_CMD} -s ${MAX_KEEP_IN_MAIL} -d ${REP_PIECE_JOINTE} ${IN
case "${JIR_TOKEN}" in case "${JIR_TOKEN}" in
"" | no | *Error* | \<* ) "" | no | *Error* | \<* )
LOG_FIC " - ${RED}Can't upload <${ATTACH_NAME}> <${ATTACH_CONTENT_TYPE}> <$(wc -c < "${ATTACH_MEDIA}")> (${JIR_TOKEN}) <in.$$.bak>. It will be not change in e-mail.${NC}" LOG_FIC " - ${RED}Can't upload <${ATTACH_NAME}> <${ATTACH_CONTENT_TYPE}> <$(wc -c < "${ATTACH_MEDIA}")> (${JIR_TOKEN}) <in.$$.bak>. It will be not change in e-mail.${NC}"
echo "new:" echo "url:"
keepFailed "${INSPECT_DIR}/in.$$" keepFailed "${INSPECT_DIR}/in.$$"
echo "UPLOAD_FAIL" >> "${JIRAFEAU_ERROR}" echo "UPLOAD_FAIL" >> "${JIRAFEAU_ERROR}"
;; ;;
@ -243,8 +240,7 @@ LOG_FIC "${CYAN}${SHRINK_CMD} -s ${MAX_KEEP_IN_MAIL} -d ${REP_PIECE_JOINTE} ${IN
LOG_FIC "${CYAN}${SHRINK_CMD} -s \"${MAX_KEEP_IN_MAIL}\" \"${INSPECT_DIR}/in.$$\" \"${INSPECT_DIR}/in.$$.altered\" 2>> \"${FIC_LOG}\"${NC}" LOG_FIC "${CYAN}${SHRINK_CMD} -s \"${MAX_KEEP_IN_MAIL}\" \"${INSPECT_DIR}/in.$$\" \"${INSPECT_DIR}/in.$$.altered\" 2>> \"${FIC_LOG}\"${NC}"
} | "${SHRINK_CMD}" -s "${MAX_KEEP_IN_MAIL}" "${INSPECT_DIR}/in.$$" "${INSPECT_DIR}/in.$$.altered" 2>> "${FIC_LOG}" } | "${SHRINK_CMD}" -s "${MAX_KEEP_IN_MAIL}" "${INSPECT_DIR}/in.$$" "${INSPECT_DIR}/in.$$.altered" 2>> "${FIC_LOG}"
# XXX trace [ -n "${DEBUG}" ] && (mkdir -p "${DIR_LOG}/pb/" ; cp "${INSPECT_DIR}/in.$$.altered" "${DIR_LOG}/pb/in.$$.altered")
# cp "${INSPECT_DIR}/in.$$" "${INSPECT_DIR}/in.$$.altered" /var/mail/tmp/
if [ -s "${JIRAFEAU_ERROR}" ]; then if [ -s "${JIRAFEAU_ERROR}" ]; then
LOG_FIC " - ${RED}upload fail${NC}" LOG_FIC " - ${RED}upload fail${NC}"

17
src/cpp/Attachment.cpp
View File

@ -65,7 +65,14 @@ const string Attachment::ALTERNATIVE ("alternative");
const regex Attachment::nameCharsetRegEx (".*name\\*=(.*)"); const regex Attachment::nameCharsetRegEx (".*name\\*=(.*)");
const regex Attachment::nameRegEx (".*name=\"([^\"]*)\".*"); const regex Attachment::nameRegEx (".*name=\"([^\"]*)\".*");
const regex Attachment::boundaryRegEx (".*boundary=\"?([^\" ]*)\"?.*"); // boundary="----=_Part_796779_1154936629.1668080348646"
// boundary="------------040709000505010508040808"
// boundary="----------=_1668606031-941125-91"
// boundary="_004_PAVPR10MB6792713B313048E3A259B215B2079PAVPR10MB6792EURP_";
// boundary="_000_PAVPR10MB6792713B313048E3A259B215B2079PAVPR10MB6792EURP_"
// boundary=--boundary_1351_64006126-2b0e-4a3b-98ac-4797d1634188
// boundary=--boundary_1352_7e294c9a-cfab-44a0-bfb3-7310380ac7cb;
const regex Attachment::boundaryRegEx (".*boundary=\"?([^\"; ]*)\"?;?.*");
const regex Attachment::cidDefRegEx (".*<([^>]*)>.*"); const regex Attachment::cidDefRegEx (".*<([^>]*)>.*");
const regex Attachment::textRegEx (".*text/("+PLAIN+"|"+HTML+").*"); const regex Attachment::textRegEx (".*text/("+PLAIN+"|"+HTML+").*");
const regex Attachment::multiRegEx ("\\s*multipart/(mixed|"+RELATED+"|"+ALTERNATIVE+").*"); const regex Attachment::multiRegEx ("\\s*multipart/(mixed|"+RELATED+"|"+ALTERNATIVE+").*");
@ -354,7 +361,11 @@ Attachment::nextBondary (ifstream &mbox, streamoff &curPos) {
++it) ++it)
if (couple.find (*it) != string::npos) { if (couple.find (*it) != string::npos) {
LOG ("find: "+ *it); LOG ("find: "+ *it);
subAttachements.back ().toUpdate = true; LOG ("size: " << subAttachements.size ());
if (subAttachements.size ())
subAttachements.back ().toUpdate = true;
else
LOG_BUG (true, continue, "eMailShrinker: bug A10: boundary format ? " << *this);
} }
prev = clearLine; prev = clearLine;
continue; continue;
@ -489,7 +500,7 @@ kaz::operator << (ostream& os, const Attachment& attachment) {
if (prop.length ()) if (prop.length ())
prop = " ["+prop+"]"; prop = " ["+prop+"]";
os << setw ((attachment.level % 20)*2) << "" << setw (10) << SizeArg (attachment.getSize ()) << " " << attachment.getContentType () os << ("****************************************"+40-(attachment.level % 20)*2) << setw (10) << SizeArg (attachment.getSize ()) << " " << attachment.getContentType ()
<< prop << (attachment.cid.length () ? " id: "+attachment.cid : "") << prop << (attachment.cid.length () ? " id: "+attachment.cid : "")
<< (attachment.boundary.length () ? " boundary: "+attachment.boundary : "") << (attachment.boundary.length () ? " boundary: "+attachment.boundary : "")
<< " (" << attachment.beginPos << " / " << attachment.contentPos << " / " << attachment.endPos << ") " << endl; << " (" << attachment.beginPos << " / " << attachment.contentPos << " / " << attachment.endPos << ") " << endl;

186
src/cpp/MainAttachment.cpp
View File

@ -32,14 +32,15 @@
// knowledge of the CeCILL-B license and that you accept its terms. // // knowledge of the CeCILL-B license and that you accept its terms. //
//////////////////////////////////////////////////////////////////////////// ////////////////////////////////////////////////////////////////////////////
#include <iostream> #include <algorithm>
#include <vector>
#include <set>
#include <fstream> #include <fstream>
#include <iomanip> #include <iomanip>
#include <iostream>
#include <math.h> #include <math.h>
#include <algorithm> #include <regex>
#include <set>
#include <unistd.h> #include <unistd.h>
#include <vector>
#include "kazDebug.hpp" #include "kazDebug.hpp"
#include "kazMisc.hpp" #include "kazMisc.hpp"
@ -50,64 +51,75 @@
using namespace std; using namespace std;
using namespace kaz; using namespace kaz;
static const string KAZ_WEB_SITE = "https://kaz.bzh/"; static const string KAZ_WEB_SITE ("https://kaz.bzh/");
static const string TMPL_DOWNLOAD = "{{DOWNLOAD}}"; static const string TMPL_DOWNLOAD ("{{DOWNLOAD}}");
static const string TMPL_FILENAME = "{{FILENAME}}"; static const string TMPL_FILENAME ("{{FILENAME}}");
static const string CID = "cid:"; static const string CID ("cid:");
static const string KAZ_PLAIN_HR = "______________________________________________________________________________"; // "l=/" => v1 compatibility
static const string KAZ_PLAIN_START = "~~ PJ-KAZ !"; // don't end whith space static const regex archiveURLSignature (".*(([&?]g=)|([&?]l=/)).*");
static const string KAZ_PLAIN_STOP = KAZ_PLAIN_START+" ~~";
static const string KAZ_PLAIN_DONT_TOUCH = "(conservez cette partie intacte dans votre réponse si vous voulez transmettre les documents précédents)";
static const string KAZ_PLAIN_WARNING = "Attention : Kaz a dépollué ce message. Les pièces jointes ont été retirées et placées dans un dépôt provisoire. Elles seront automatiquement supprimées dans 1 mois. Si elles sont importantes et que vous souhaitez les conserver, vous devez utiliser les liens ci-dessous. Pour mieux comprendre la politique de nos services visitez kaz.bzh";
static const string KAZ_PLAIN_DOWLOAD_ONE = "Vos pièces jointes sont à télécharger individuellement ici :";
static const string KAZ_PLAIN_DOWLOAD_OTHER = "(Contenu dans des messages précédents)";
static const string KAZ_PLAIN_DOWLOAD_ALL = "Vous pouvez télécharger l'ensemble dans une archive là :";
static const string HEAD = "<head>";
static const string HEAD_END = "</head>";
static const string KAZ_CSS_URL = "https://kaz.bzh/m/email.css";
static const string KAZ_CSS = "<link rel=\"stylesheet\" type=\"text/css\" charset=\"utf-8\" href=\""+KAZ_CSS_URL+"\"/>";
static const string A_END = "</a>";
static const string LI_BEGIN = "<li";
static const string CLASS_ONE = "class=\"one\"";
static const string LI_ONE = LI_BEGIN+" "+CLASS_ONE+">";
static const string LI_ALL = LI_BEGIN+" class=\"all\">";
static const string LI_END = "</li>";
static const string HREF_ONE = "href=\"";
static const string BODY_END = "</body>";
static const string HTML_END = "</html>";
static const string KAZ_HTML_TAG = "<!--KAZ"; // don't end whith space static const string KAZ_PLAIN_HR ("______________________________________________________________________________");
static const string KAZ_HTML_START = KAZ_HTML_TAG+" START-->"; static const string KAZ_PLAIN_START ("~~ PJ-KAZ !"); // don't end whith space
static const string KAZ_HTML_STOP = KAZ_HTML_TAG+" STOP-->"; static const string KAZ_PLAIN_STOP (KAZ_PLAIN_START+" ~~");
static const string KAZ_PLAIN_DONT_TOUCH ("(conservez cette partie intacte dans votre réponse si vous voulez transmettre les documents précédents)");
static const string KAZ_PLAIN_WARNING ("Attention : Kaz a dépollué ce message. Les pièces jointes ont été retirées et placées dans un dépôt provisoire. "
"Elles seront automatiquement supprimées dans 1 mois. "
"Si elles sont importantes et que vous souhaitez les conserver, vous devez utiliser les liens ci-dessous. "
"Pour mieux comprendre la politique de nos services visitez kaz.bzh");
static const string KAZ_PLAIN_DOWLOAD_ONE ("Vos pièces jointes sont à télécharger individuellement ici :");
static const string KAZ_PLAIN_DOWLOAD_OTHER ("(Contenu dans des messages précédents)");
static const string KAZ_PLAIN_DOWLOAD_ALL ("Vous pouvez télécharger l'ensemble dans une archive là :");
static const string HEAD ("<head>");
static const string HEAD_END ("</head>");
static const string KAZ_CSS_URL ("https://kaz.bzh/m/email.css");
static const string KAZ_CSS ("<link rel=\"stylesheet\" type=\"text/css\" charset=\"utf-8\" href=\""+KAZ_CSS_URL+"\"/>");
static const string A_END ("</a>");
static const string LI_BEGIN ("<li");
static const string CLASS_ONE ("class=\"one\"");
static const string LI_ONE (LI_BEGIN+" "+CLASS_ONE+">");
static const string LI_ALL (LI_BEGIN+" class=\"all\">");
static const string LI_END ("</li>");
static const string HREF_ONE ("href=\"");
static const string BODY_END ("</body>");
static const string HTML_END ("</html>");
static const string KAZ_HTML_TAG ("<!--KAZ"); // don't end whith space
static const string KAZ_HTML_START (KAZ_HTML_TAG+" START-->");
static const string KAZ_HTML_STOP (KAZ_HTML_TAG+" STOP-->");
// Textes précédents encodés en SGML // Textes précédents encodés en SGML
static const string KAZ_HTML_DONT_TOUCH = "(conservez cette partie intacte dans votre r&eacute;ponse si vous voulez transmettre les documents pr&eacute;c&eacute;dents)"; static const string KAZ_HTML_DONT_TOUCH ("(conservez cette partie intacte dans votre r&eacute;ponse si vous voulez transmettre les documents pr&eacute;c&eacute;dents)");
static const string KAZ_HTML_DOWLOAD_ONE = "Vos pi&egrave;ces jointes sont &agrave; t&eacute;l&eacute;charger individuellement ici :"; static const string KAZ_HTML_DOWLOAD_ONE ("Vos pi&egrave;ces jointes sont &agrave; t&eacute;l&eacute;charger individuellement ici :");
static const string KAZ_HTML_DOWLOAD_OTHER = "(Contenu dans des messages pr&eacute;c&eacute;dents)"; static const string KAZ_HTML_DOWLOAD_OTHER ("(Contenu dans des messages pr&eacute;c&eacute;dents)");
static const string KAZ_HTML_DOWLOAD_ALL = "Vous pouvez t&eacute;l&eacute;charger l'ensemble dans une archive l&agrave; :"; static const string KAZ_HTML_DOWLOAD_ALL ("Vous pouvez t&eacute;l&eacute;charger l'ensemble dans une archive l&agrave; :");
static const string KAZ_HTML_ARCHIVE = "archive"; static const string KAZ_HTML_ARCHIVE ("archive");
static const string KAZ_EMPTY_TEXT_PLAIN ("Content-Type: text/plain; charset=utf-8\n"
"Content-Transfer-Encoding: base64\n");
// ================================================================================ // ================================================================================
vector <string> vector <string>
Attachment::stringsToUpdate ({KAZ_PLAIN_START, "\""+CID}); Attachment::stringsToUpdate ({KAZ_PLAIN_START, "\""+CID});
// ================================================================================ // ================================================================================
const string MainAttachment::templatePlainAddLink (" - "+TMPL_DOWNLOAD+" "+TMPL_FILENAME+"\r\n"); const string MainAttachment::templatePlainAddLink (" * "+TMPL_FILENAME+" <"+TMPL_DOWNLOAD+">\r\n");
const string MainAttachment::templatePlainAllLink ("\r\n"+KAZ_PLAIN_DOWLOAD_ALL+"\r\n * "+TMPL_DOWNLOAD+"\r\n"); const string MainAttachment::templatePlainAllLink ("\r\n * "+KAZ_PLAIN_DOWLOAD_ALL+" <"+TMPL_DOWNLOAD+">\r\n");
const string MainAttachment::templateHtmlHeader (KAZ_HTML_START+"<p style=\"clear: left; padding: 1pc 0 0 0; font-size:10px; color:#969696;\">"+KAZ_PLAIN_START+"</p><hr>\n" const string MainAttachment::templateHtmlHeader (KAZ_HTML_START+"<p style=\"clear: left; padding: 1pc 0 0 0; font-size:10px; color:#969696;\">"+KAZ_PLAIN_START+"</p><hr>\n"
"<div class=\"kaz\">" "<div class=\"kaz\">"
"<p style=\"font-size:10px; color:#969696;\">"+KAZ_HTML_DONT_TOUCH+"</p>\n" "<p style=\"font-size:10px; color:#969696;\">"+KAZ_HTML_DONT_TOUCH+"</p>\n"
"<p>"+KAZ_HTML_DOWLOAD_ONE+"<ul>\n"); "<p>"+KAZ_HTML_DOWLOAD_ONE+"<ul>\n");
const string MainAttachment::templateHtmlAddLink (LI_ONE+"<a "+HREF_ONE+TMPL_DOWNLOAD+"\">"+TMPL_FILENAME+"</a>"+LI_END+"\n"); const string MainAttachment::templateHtmlAddLink (LI_ONE+"<a "+HREF_ONE+TMPL_DOWNLOAD+"\">"+TMPL_FILENAME+"</a>"+LI_END+"\n");
const string MainAttachment::templateHtmlOtherLink ("</ul>"+KAZ_HTML_DOWLOAD_OTHER+"<ul>\n"); const string MainAttachment::templateHtmlOtherLink ("</ul>"+KAZ_HTML_DOWLOAD_OTHER+"<ul>\n");
const string MainAttachment::templateHtmlAllLink ("</ul><ul>"+LI_ALL+KAZ_HTML_DOWLOAD_ALL+" <a href=\""+TMPL_DOWNLOAD+"\">"+KAZ_HTML_ARCHIVE+"</a>"+LI_END+"\n"); const string MainAttachment::templateHtmlAllLink ("</ul><ul>"+LI_ALL+KAZ_HTML_DOWLOAD_ALL+" <a href=\""+TMPL_DOWNLOAD+"\">"+KAZ_HTML_ARCHIVE+"</a>"+LI_END+"\n");
const string MainAttachment::templateHtmlFooter ("</ul></p>\n" const string MainAttachment::templateHtmlFooter ("</ul></p>\n"
"<p class=\"msg\"><a class=\"kaz\" href=\""+KAZ_WEB_SITE+"\"> "+KAZ_WEB_SITE+" </a></p></div>\n" "<p class=\"msg\"><a class=\"kaz\" href=\""+KAZ_WEB_SITE+"\"> "+KAZ_WEB_SITE+" </a></p></div>\n"
"<hr><p style=\"font-size:10px; color:#969696;\">"+KAZ_PLAIN_STOP+"</p>"+KAZ_HTML_STOP+"\n"); "<hr><p style=\"font-size:10px; color:#969696;\">"+KAZ_PLAIN_STOP+"</p>"+KAZ_HTML_STOP+"\n");
const regex MainAttachment::whiteSpaceRegEx ("\\s+"); const regex MainAttachment::whiteSpaceRegEx ("\\s+");
// ================================================================================ // ================================================================================
void void
@ -167,7 +179,8 @@ MainAttachment::addLink (string &plain, string &html, const string &url, const s
plain += plainNewOneLink; plain += plainNewOneLink;
string htmlNewOneLink (templateHtmlAddLink); string htmlNewOneLink (templateHtmlAddLink);
string codedUrl (url); string codedUrl (url);
replaceAll (codedUrl, "&", "&amp;"); // XXX amp ?
//replaceAll (codedUrl, "&", "&amp;");
replaceAll (htmlNewOneLink, TMPL_DOWNLOAD, codedUrl); replaceAll (htmlNewOneLink, TMPL_DOWNLOAD, codedUrl);
replaceAll (htmlNewOneLink, TMPL_FILENAME, name); replaceAll (htmlNewOneLink, TMPL_FILENAME, name);
html += htmlNewOneLink; html += htmlNewOneLink;
@ -218,11 +231,11 @@ MainAttachment::getDisclaim (string &plain, string &html) const {
string allPlainLinks (templatePlainAllLink); string allPlainLinks (templatePlainAllLink);
replaceAll (allPlainLinks, TMPL_DOWNLOAD, archiveDownloadURL); replaceAll (allPlainLinks, TMPL_DOWNLOAD, archiveDownloadURL);
plain += allPlainLinks; plain += allPlainLinks;
string allLinks (templateHtmlAllLink); string allHtmlLinks (templateHtmlAllLink);
replaceAll (allLinks, TMPL_DOWNLOAD, archiveDownloadURL); replaceAll (allHtmlLinks, TMPL_DOWNLOAD, archiveDownloadURL);
html += allLinks; html += allHtmlLinks;
} }
html += templateHtmlFooter; html += templateHtmlFooter+"\r\n";
plain += "\r\n\r\n"+KAZ_WEB_SITE+"\r\n"+KAZ_PLAIN_HR+"\r\n"+KAZ_PLAIN_STOP+"\r\n"; plain += "\r\n\r\n"+KAZ_WEB_SITE+"\r\n"+KAZ_PLAIN_HR+"\r\n"+KAZ_PLAIN_STOP+"\r\n";
// & => &amp; done // & => &amp; done
LOG ("plain: " << plain); LOG ("plain: " << plain);
@ -245,6 +258,7 @@ MainAttachment::addPrevious (const string &href, const string &name, const bool
void void
MainAttachment::extractLinks (const string &extractedPlainKAZ) { MainAttachment::extractLinks (const string &extractedPlainKAZ) {
// plain text => "* name <url>"
DEF_LOG ("Attachment::extractedPlainKAZ", "extractedPlainKAZ: " << extractedPlainKAZ); DEF_LOG ("Attachment::extractedPlainKAZ", "extractedPlainKAZ: " << extractedPlainKAZ);
for (string::size_type startPos (0); for (string::size_type startPos (0);
(startPos = extractedPlainKAZ.find ("http", startPos)) != string::npos; (startPos = extractedPlainKAZ.find ("http", startPos)) != string::npos;
@ -257,22 +271,30 @@ MainAttachment::extractLinks (const string &extractedPlainKAZ) {
if (extractedPlainKAZ [stopPos]) if (extractedPlainKAZ [stopPos])
++stopPos; ++stopPos;
streamoff stopName = startPos;
startPos = stopPos; startPos = stopPos;
// get all href but KAZ_WEB_SITE // get all href but KAZ_WEB_SITE
// the archive link will be skip by filter.sh // the archive link will be skip by filter.sh
if (href == KAZ_WEB_SITE) if (href == KAZ_WEB_SITE)
continue; continue;
string::size_type nextPos = extractedPlainKAZ.find ("http", startPos); // backward find name
string name (extractedPlainKAZ, startPos, (nextPos == string::npos ? extractedPlainKAZ.length () : nextPos) - startPos); string::size_type startName = extractedPlainKAZ.rfind ("* ", stopName);
// skip [> \r\n\t] string name;
nextPos = name.find_first_not_of ("[> \t\r\n\"]");
if (nextPos != string::npos) if (startName != string::npos) {
name.erase (0, nextPos); name = string (extractedPlainKAZ, startName+3, stopName - startName - 3);
// stop before [>\r\n\t] // skip [> \r\n\t]
nextPos = name.find_first_of ("[\"]"); string::size_type nextPos = name.find_first_not_of ("[>< \t\r\n\"]");
if (nextPos != string::npos) if (nextPos != string::npos)
name.erase (nextPos); name.erase (0, nextPos);
// stop before [>\r\n\t]
nextPos = name.find_first_of ("[\"<]");
if (nextPos != string::npos)
name.erase (nextPos);
}
LOG ("plain name: " << name); LOG ("plain name: " << name);
addPrevious (href, name); addPrevious (href, name);
} }
@ -281,6 +303,7 @@ MainAttachment::extractLinks (const string &extractedPlainKAZ) {
// ================================================================================ // ================================================================================
void void
MainAttachment::extractLinks (const vector<string> &liOne) { MainAttachment::extractLinks (const vector<string> &liOne) {
// html text => "<li ...><a href="url">name</a>"
DEF_LOG ("Attachment::extractedPlainKAZ", "liOne.size: " << liOne.size ()); DEF_LOG ("Attachment::extractedPlainKAZ", "liOne.size: " << liOne.size ());
for (const string &one : liOne) { for (const string &one : liOne) {
if (caseInsensitiveFind (one, CLASS_ONE) == string::npos) if (caseInsensitiveFind (one, CLASS_ONE) == string::npos)
@ -353,17 +376,19 @@ MainAttachment::extractPreviousKAZ (ifstream &mbox) {
void void
MainAttachment::removePreviousArchive () { MainAttachment::removePreviousArchive () {
vector<string> toRemove; vector<string> toRemove;
for (map <string, string>::const_iterator it = previousLinks.begin (); it != previousLinks.end (); ++it) for (map <string, string>::const_iterator it = previousLinks.begin (); it != previousLinks.end (); ++it) {
if (it->first.find ("&g=") != string::npos || const string key (it->first);
it->first.find ("&l=/") != string::npos) // v1 compatibility if (regex_match (key, archiveURLSignature))
toRemove.push_back (it->first); toRemove.push_back (key);
}
for (string old : toRemove) for (string old : toRemove)
previousLinks.erase (old); previousLinks.erase (old);
} }
// ================================================================================ // ================================================================================
MainAttachment::MainAttachment (ifstream &mbox) MainAttachment::MainAttachment (ifstream &mbox)
: Attachment (mbox, initTmpLevel (), 0, initTmpPos ()) { : Attachment (mbox, initTmpLevel (), 0, initTmpPos ()),
forceMainText (false) {
DEF_LOG ("MainAttachment::MainAttachment", ""); DEF_LOG ("MainAttachment::MainAttachment", "");
string line; string line;
for (; getline (mbox, line); ) for (; getline (mbox, line); )
@ -377,6 +402,7 @@ MainAttachment::markSignificant (const streamoff &minAttachSize, ifstream &mbox)
DEF_LOG ("MainAttachment::markSignificant", "minAttachSize: " << minAttachSize); DEF_LOG ("MainAttachment::markSignificant", "minAttachSize: " << minAttachSize);
bool plainMarked (false), htmlMarked (false); bool plainMarked (false), htmlMarked (false);
markDisclaim (plainMarked, htmlMarked); markDisclaim (plainMarked, htmlMarked);
forceMainText = ! (plainMarked || htmlMarked);
Attachment::markSignificant ("", minAttachSize, mbox, allMarkedPtrs); Attachment::markSignificant ("", minAttachSize, mbox, allMarkedPtrs);
} }
@ -512,7 +538,7 @@ MainAttachment::substitute (ifstream &mbox, ofstream &outbox, const SizeArg &min
} }
if (attachP->cid.length ()) { if (attachP->cid.length ()) {
string tmp (attachP->downloadUrl); string tmp (attachP->downloadUrl);
replaceAll (tmp, "&", "&amp;"); //replaceAll (tmp, "&", "&amp;");
translateHtml.insert (pair<const string, const string> (CID+attachP->cid, tmp)); translateHtml.insert (pair<const string, const string> (CID+attachP->cid, tmp));
} }
} }
@ -521,15 +547,31 @@ MainAttachment::substitute (ifstream &mbox, ofstream &outbox, const SizeArg &min
continue; continue;
for (EmbeddedData &embedded : attachP->embeddedData) for (EmbeddedData &embedded : attachP->embeddedData)
readDownloadUrl (embedded.downloadUrl); readDownloadUrl (embedded.downloadUrl);
// XXX test empty ?
} }
readArchiveUrl (); readArchiveUrl ();
removePreviousArchive ();
string plainDisclaim, htmlDisclaim; string plainDisclaim, htmlDisclaim;
getDisclaim (plainDisclaim, htmlDisclaim); getDisclaim (plainDisclaim, htmlDisclaim);
// copy email // copy email
streamoff curPos = 0; streamoff curPos = 0;
if (forceMainText) {
cerr << endl << endl << " #################### coucou " << forceMainText << " " << contentPos << " " << *this << endl;
// check no main text
LOG ("Force main text");
LOG_BUG (boundary.empty () || ! subAttachements.size (), /**/, "eMailShrinker: can't force add footer M9: : " << *this);
copy (mbox, outbox, curPos, contentPos);
curPos = contentPos;
cerr << " #################### coucou " << curPos << endl << endl;
string content (plainDisclaim);
base64Encode (content);
outbox << boundary.substr (0, boundary.length () -2) << endl
<< KAZ_EMPTY_TEXT_PLAIN << endl
<< content << endl;
outbox.flush ();
}
for (Attachment *attachP : allMarkedPtrs) { for (Attachment *attachP : allMarkedPtrs) {
copy (mbox, outbox, curPos, attachP->beginInParent); copy (mbox, outbox, curPos, attachP->beginInParent);
LOG_BUG (attachP->toUpdate && attachP->toExtract, /**/, "eMailShrinker: bug M5: update and extract. pos: " << attachP->beginPos); LOG_BUG (attachP->toUpdate && attachP->toExtract, /**/, "eMailShrinker: bug M5: update and extract. pos: " << attachP->beginPos);
if (attachP->toExtract) { if (attachP->toExtract) {

10
src/cpp/eMailShrinker.cpp
View File

@ -32,7 +32,7 @@
// knowledge of the CeCILL-B license and that you accept its terms. // // knowledge of the CeCILL-B license and that you accept its terms. //
//////////////////////////////////////////////////////////////////////////// ////////////////////////////////////////////////////////////////////////////
#define LAST_VERSION "2.0 2022-02-08 eMailShrinker" #define LAST_VERSION "2.1 2022-10-30 eMailShrinker"
#include <iostream> #include <iostream>
#include <fstream> #include <fstream>
@ -188,10 +188,10 @@ main (int argc, char** argv) {
MainAttachment attachment (mbox); MainAttachment attachment (mbox);
mbox.close (); mbox.close ();
if (attachment.getBoundary ().empty ()) { // if (attachment.getBoundary ().empty ()) {
cerr << "no attachment" << endl; // cerr << "no attachment" << endl;
return 1; // return 1;
} // }
// parse structure // parse structure
mbox.open (inputName); mbox.open (inputName);
attachment.markSignificant (minAttachSize, mbox); attachment.markSignificant (minAttachSize, mbox);

7
src/cpp/jirafeauAPI.cpp
View File

@ -32,7 +32,7 @@
// knowledge of the CeCILL-B license and that you accept its terms. // // knowledge of the CeCILL-B license and that you accept its terms. //
//////////////////////////////////////////////////////////////////////////// ////////////////////////////////////////////////////////////////////////////
#define LAST_VERSION "1.0 2021-02-21 jirafeauAPI" #define LAST_VERSION "1.1 2022-10-30 jirafeauAPI"
#include <iostream> #include <iostream>
#include <string> #include <string>
@ -281,12 +281,13 @@ main (int argc, char** argv) {
} }
curl_easy_setopt (easyhandle, CURLOPT_MIMEPOST, multipart); curl_easy_setopt (easyhandle, CURLOPT_MIMEPOST, multipart);
curl_easy_perform (easyhandle); CURLcode res (curl_easy_perform (easyhandle));
curl_easy_cleanup (easyhandle); curl_easy_cleanup (easyhandle);
cout << readBuffer << endl; cout << readBuffer << endl;
showTime ("Upload"); showTime ("Upload");
if (res != CURLE_OK)
cerr << prog << " failed: " << curl_easy_strerror (res) << endl;
return 0; return 0;
} }

4
src/include/Attachment.hpp
View File

@ -35,9 +35,9 @@
#ifndef _kaz_Attachment_hpp #ifndef _kaz_Attachment_hpp
#define _kaz_Attachment_hpp #define _kaz_Attachment_hpp
#include <string>
#include <regex>
#include <map> #include <map>
#include <regex>
#include <string>
#include <utility> #include <utility>
#include "EmbeddedData.hpp" #include "EmbeddedData.hpp"

2
src/include/MainAttachment.hpp
View File

@ -83,6 +83,8 @@ namespace kaz {
bfs::path extractDir; bfs::path extractDir;
/*! URL for download archives */ /*! URL for download archives */
string archiveDownloadURL; string archiveDownloadURL;
/*! no main text in email can be use to add disclaim */
bool forceMainText;
/*! subset in the tree of all attachments to be consider for extraction or modification */ /*! subset in the tree of all attachments to be consider for extraction or modification */
vector<Attachment *> allMarkedPtrs; vector<Attachment *> allMarkedPtrs;