From b4ea2185f0b1c25a4b15f835d6781936d532ffd6 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Fran=C3=A7ois?= <francois@kaz.bzh>
Date: Mon, 31 Oct 2022 07:08:19 +0100
Subject: [PATCH] fix synchro plain and html

---
 README.md                  | 11 +++++++
 src/Jirafeau/a.php         | 26 ++++++++++++---
 src/bash/filter.sh         | 20 +++++-------
 src/cpp/MainAttachment.cpp | 65 ++++++++++++++++++++++++--------------
 src/cpp/eMailShrinker.cpp  | 10 +++---
 src/cpp/jirafeauAPI.cpp    |  7 ++--
 src/include/Attachment.hpp |  4 +--
 7 files changed, 94 insertions(+), 49 deletions(-)

diff --git a/README.md b/README.md
index 1affbd3..869461b 100644
--- a/README.md
+++ b/README.md
@@ -41,3 +41,14 @@ git clone https://git.kaz.bzh/KAZ/depollueur.git
 
 make -j $(nproc)
 ```
+
+##Astuces
+
+
+Pour la mise au point dans Kaz, il faut bloquer la mise à jour automatique avec :
+
+``` bash
+touch /kaz/git/depollueur/dontupdage
+/kaz/bin/container.sh stop jirafeau postfix; docker system prune; /kaz/bin/install.sh jirafeau postfix
+```
+
diff --git a/src/Jirafeau/a.php b/src/Jirafeau/a.php
index 736b1fa..26f4c72 100644
--- a/src/Jirafeau/a.php
+++ b/src/Jirafeau/a.php
@@ -12,6 +12,7 @@
  a.php?s=mel@domain.org&t=password + [action] => manage account
  action: a=login a=logout a=r[on|off] a=p[minute|hour|day|week|month|quarter]
 */
+
 use PHPMailer\PHPMailer\PHPMailer;
 use PHPMailer\PHPMailer\SMTP;
 use PHPMailer\PHPMailer\Exception;
@@ -748,10 +749,17 @@ if ($doDownload) {
                     $modif = true;
                     continue;
                 }
-                $srcName = $dstName = ($link ['file_name'] ? $link ['file_name'] : M_NO_FILENAME);
-                if (in_array ($srcName, $singleName))
-                    for ($i = 0; $i < 10000; ++$i) {
-                        $dstName = sprintf ("%s-%2d", $srcName, $i);
+                $dstName = ($link ['file_name'] ? $link ['file_name'] : M_NO_FILENAME);
+                if (in_array ($dstName, $singleName)) {
+                    $dstFilename = $dstName;
+                    $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, $singleName)) {
                             if (isset ($archiveContent [T_RENAME]))
                                 ++$archiveContent [T_RENAME];
@@ -761,6 +769,7 @@ if ($doDownload) {
                             break;
                         }
                     }
+                }
                 $singleName [] = $dstName;
                 $archiveContent [$cat][T_ENTRIES][] = [T_HASH => $link ['hash'], T_FILENAME => $dstName, T_CRYPT_KEY => $cryptKey, T_CRYPTED => $link ['crypted']];
             }
@@ -1080,6 +1089,15 @@ echo
     '<select name="'.A_ACTION.'" style="width: auto !important;">';
 foreach ($periodText as $item => $text)
     echo '  <option value="'.A_PERIOD.$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
     '</select> '.
     '<button type="submit">'."valider".'</button>'.
diff --git a/src/bash/filter.sh b/src/bash/filter.sh
index 60936a8..0920792 100644
--- a/src/bash/filter.sh
+++ b/src/bash/filter.sh
@@ -66,6 +66,7 @@ ARCHIVE_TITLE="archive_content"
 ARCHIVE_MIME="text/kaz_email_archive"
 
 KEEP_FAILED=true
+DEBUG=
 
 ####################  FONCTIONS ############################################
 BOLD=''
@@ -134,7 +135,7 @@ ARCHIVE_CONTENT="${REP_PIECE_JOINTE}/archive-content.txt"
 JIRAFEAU_ERROR="${REP_PIECE_JOINTE}/jirafeau-error.txt"
 
 # 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.$$"; then
     LOG_FIC "${RED}Cannot save mail to file${NC}"
@@ -142,8 +143,7 @@ if ! cat > "in.$$"; then
 fi
 LOG_FIC "\n" \
 	"             size: ${YELLOW}$(wc -c < "${INSPECT_DIR}/in.$$")${NC}"
-# XXX trace
-# cp "${INSPECT_DIR}/in.$$" "${INSPECT_DIR}/in.$$.bak"
+[ -n "${DEBUG}" ] && (mkdir -p "${DIR_LOG}/pb/" ; cp "${INSPECT_DIR}/in.$$" "${DIR_LOG}/pb/in.$$.orig")
 
 mkdir -p "${REP_PIECE_JOINTE}/"
 >"${OLD_LINKS}"
@@ -156,8 +156,7 @@ echo "time: ${DATE_TEMPS}\nid: $(date +%s)" > "${ARCHIVE_CONTENT}"
 LOG_FIC "${CYAN}${SHRINK_CMD} -u \"${INSPECT_DIR}/in.$$\" 2>> \"${FIC_LOG}\" > \"${OLD_LINKS}\"${NC}"
 "${SHRINK_CMD}" -u "${INSPECT_DIR}/in.$$" 2>> "${FIC_LOG}" > "${OLD_LINKS}"
 
-cat "${OLD_LINKS}" | grep "${JIRAFEAU_URL}" | while read REMOTE_LINK
-do
+cat "${OLD_LINKS}" | grep "${JIRAFEAU_URL}" | while read REMOTE_LINK; do
     REMOTE_REF=$(echo "${REMOTE_LINK}" | sed -e 's/.*h=\([^&]*\).*/\1/' -e 's/.*http.*//')
     [ -z "${REMOTE_REF}" ] && continue
     REMOTE_KEY=$(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
 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>> "${FIC_LOG}" | {
-    while read ATTACH_TMP_NAME
-    do
-	if [ -d "${ATTACH_TMP_NAME}" ]
-	then
+    while read ATTACH_TMP_NAME; do
+	if [ -d "${ATTACH_TMP_NAME}" ]; then
 	    ATTACH_MEDIA="${ATTACH_TMP_NAME}/media"
 	    ATTACH_NAME=$(grep "^Name: " "${ATTACH_TMP_NAME}/meta" | cut -c 7- )
 	    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
 		"" | no | *Error* | \<* )
 		    LOG_FIC " - ${RED}Can't upload <${ATTACH_NAME}> <${ATTACH_CONTENT_TYPE}> <$(wc -c < "${ATTACH_MEDIA}")> (${JIR_TOKEN}) <in.$$.bak>. It will be not change in e-mail.${NC}"
-		    echo "new:"
+		    echo "url:"
 		    keepFailed "${INSPECT_DIR}/in.$$"
 		    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}"
 } | "${SHRINK_CMD}" -s "${MAX_KEEP_IN_MAIL}" "${INSPECT_DIR}/in.$$" "${INSPECT_DIR}/in.$$.altered" 2>> "${FIC_LOG}"
 
-# XXX trace
-# cp "${INSPECT_DIR}/in.$$" "${INSPECT_DIR}/in.$$.altered" /var/mail/tmp/
+[ -n "${DEBUG}" ] && (mkdir -p "${DIR_LOG}/pb/" ; cp "${INSPECT_DIR}/in.$$.altered" "${DIR_LOG}/pb/in.$$.altered")
 
 if [ -s "${JIRAFEAU_ERROR}" ]; then
     LOG_FIC " - ${RED}upload fail${NC}"
diff --git a/src/cpp/MainAttachment.cpp b/src/cpp/MainAttachment.cpp
index 2e49f75..0258633 100644
--- a/src/cpp/MainAttachment.cpp
+++ b/src/cpp/MainAttachment.cpp
@@ -32,14 +32,15 @@
 // knowledge of the CeCILL-B license and that you accept its terms.	  //
 ////////////////////////////////////////////////////////////////////////////
 
-#include <iostream>
-#include <vector>
-#include <set>
+#include <algorithm>
 #include <fstream>
 #include <iomanip>
+#include <iostream>
 #include <math.h>
-#include <algorithm>
+#include <regex>
+#include <set>
 #include <unistd.h>
+#include <vector>
 
 #include "kazDebug.hpp"
 #include "kazMisc.hpp"
@@ -55,6 +56,10 @@ static const string TMPL_DOWNLOAD		= "{{DOWNLOAD}}";
 static const string TMPL_FILENAME		= "{{FILENAME}}";
 static const string CID				= "cid:";
 
+// "l=/" => v1 compatibility
+static const regex archiveURLSignature		(".*(([&?]g=)|([&?]l=/)).*");
+
+
 static const string KAZ_PLAIN_HR		= "______________________________________________________________________________";
 static const string KAZ_PLAIN_START		= "~~ PJ-KAZ !"; // don't end whith space
 static const string KAZ_PLAIN_STOP		= KAZ_PLAIN_START+" ~~";
@@ -93,8 +98,8 @@ vector <string>
 Attachment::stringsToUpdate ({KAZ_PLAIN_START, "\""+CID});
 
 // ================================================================================
-const string MainAttachment::templatePlainAddLink (" - "+TMPL_DOWNLOAD+" "+TMPL_FILENAME+"\r\n");
-const string MainAttachment::templatePlainAllLink ("\r\n"+KAZ_PLAIN_DOWLOAD_ALL+"\r\n * "+TMPL_DOWNLOAD+"\r\n");
+const string MainAttachment::templatePlainAddLink ("  * "+TMPL_FILENAME+" <"+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"
 						 "<div class=\"kaz\">"
@@ -167,6 +172,7 @@ MainAttachment::addLink (string &plain, string &html, const string &url, const s
   plain += plainNewOneLink;
   string htmlNewOneLink (templateHtmlAddLink);
   string codedUrl (url);
+  // XXX amp ?
   replaceAll (codedUrl, "&", "&amp;");
   replaceAll (htmlNewOneLink, TMPL_DOWNLOAD, codedUrl);
   replaceAll (htmlNewOneLink, TMPL_FILENAME, name);
@@ -218,9 +224,9 @@ MainAttachment::getDisclaim (string &plain, string &html) const {
     string allPlainLinks (templatePlainAllLink);
     replaceAll (allPlainLinks, TMPL_DOWNLOAD, archiveDownloadURL);
     plain += allPlainLinks;
-    string allLinks (templateHtmlAllLink);
-    replaceAll (allLinks, TMPL_DOWNLOAD, archiveDownloadURL);
-    html += allLinks;
+    string allHtmlLinks (templateHtmlAllLink);
+    replaceAll (allHtmlLinks, TMPL_DOWNLOAD, archiveDownloadURL);
+    html += allHtmlLinks;
   }
   html += templateHtmlFooter;
   plain += "\r\n\r\n"+KAZ_WEB_SITE+"\r\n"+KAZ_PLAIN_HR+"\r\n"+KAZ_PLAIN_STOP+"\r\n";
@@ -245,6 +251,7 @@ MainAttachment::addPrevious (const string &href, const string &name, const bool
 
 void
 MainAttachment::extractLinks (const string &extractedPlainKAZ) {
+  // plain text => "* name <url>"
   DEF_LOG ("Attachment::extractedPlainKAZ", "extractedPlainKAZ: " << extractedPlainKAZ);
   for (string::size_type startPos (0);
        (startPos = extractedPlainKAZ.find ("http", startPos)) != string::npos;
@@ -257,22 +264,30 @@ MainAttachment::extractLinks (const string &extractedPlainKAZ) {
 
     if (extractedPlainKAZ [stopPos])
       ++stopPos;
+    streamoff stopName = startPos;
     startPos = stopPos;
+
     // get all href but KAZ_WEB_SITE
     // the archive link will be skip by filter.sh
     if (href == KAZ_WEB_SITE)
       continue;
 
-    string::size_type nextPos = extractedPlainKAZ.find ("http", startPos);
-    string name (extractedPlainKAZ, startPos, (nextPos == string::npos ? extractedPlainKAZ.length () : nextPos) - startPos);
-    // skip [> \r\n\t]
-    nextPos = name.find_first_not_of ("[> \t\r\n\"]");
-    if (nextPos != string::npos)
-      name.erase (0, nextPos);
-    // stop before [>\r\n\t]
-    nextPos = name.find_first_of ("[\"]");
-    if (nextPos != string::npos)
-      name.erase (nextPos);
+    // backward find name
+    string::size_type startName = extractedPlainKAZ.rfind ("* ", stopName);
+    string name;
+
+    if (startName != string::npos) {
+      name = string (extractedPlainKAZ, startName+3, stopName - startName - 3);
+      // skip [> \r\n\t]
+      string::size_type nextPos = name.find_first_not_of ("[>< \t\r\n\"]");
+      if (nextPos != string::npos)
+	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);
     addPrevious (href, name);
   }
@@ -281,6 +296,7 @@ MainAttachment::extractLinks (const string &extractedPlainKAZ) {
 // ================================================================================
 void
 MainAttachment::extractLinks (const vector<string> &liOne) {
+  // html text => "<li ...><a href="url">name</a>"
   DEF_LOG ("Attachment::extractedPlainKAZ", "liOne.size: " << liOne.size ());
   for (const string &one : liOne) {
     if (caseInsensitiveFind (one, CLASS_ONE) == string::npos)
@@ -353,10 +369,11 @@ MainAttachment::extractPreviousKAZ (ifstream &mbox) {
 void
 MainAttachment::removePreviousArchive () {
   vector<string> toRemove;
-  for (map <string, string>::const_iterator it = previousLinks.begin (); it != previousLinks.end (); ++it)
-    if (it->first.find ("&g=") != string::npos ||
-	it->first.find ("&l=/") != string::npos) // v1 compatibility
-      toRemove.push_back (it->first);
+  for (map <string, string>::const_iterator it = previousLinks.begin (); it != previousLinks.end (); ++it) {
+    const string key (it->first);
+    if (regex_match (key, archiveURLSignature))
+      toRemove.push_back (key);
+  }
   for (string old : toRemove)
     previousLinks.erase (old);
 }
@@ -521,8 +538,10 @@ MainAttachment::substitute (ifstream &mbox, ofstream &outbox, const SizeArg &min
       continue;
     for (EmbeddedData &embedded : attachP->embeddedData)
       readDownloadUrl (embedded.downloadUrl);
+    // XXX test empty ?
   }
   readArchiveUrl ();
+  removePreviousArchive ();
   string plainDisclaim, htmlDisclaim;
   getDisclaim (plainDisclaim, htmlDisclaim);
   // copy email
diff --git a/src/cpp/eMailShrinker.cpp b/src/cpp/eMailShrinker.cpp
index 6eb094a..5340bdd 100644
--- a/src/cpp/eMailShrinker.cpp
+++ b/src/cpp/eMailShrinker.cpp
@@ -32,7 +32,7 @@
 // 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 <fstream>
@@ -188,10 +188,10 @@ main (int argc, char** argv) {
   MainAttachment attachment (mbox);
   mbox.close ();
 
-  if (attachment.getBoundary ().empty ()) {
-    cerr << "no attachment" << endl;
-    return 1;
-  }
+  // if (attachment.getBoundary ().empty ()) {
+  //   cerr << "no attachment" << endl;
+  //   return 1;
+  // }
   // parse structure
   mbox.open (inputName);
   attachment.markSignificant (minAttachSize, mbox);
diff --git a/src/cpp/jirafeauAPI.cpp b/src/cpp/jirafeauAPI.cpp
index 17b6470..46b0706 100644
--- a/src/cpp/jirafeauAPI.cpp
+++ b/src/cpp/jirafeauAPI.cpp
@@ -32,7 +32,7 @@
 // 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 <string>
@@ -281,12 +281,13 @@ main (int argc, char** argv) {
   }
 
   curl_easy_setopt (easyhandle, CURLOPT_MIMEPOST, multipart);
-  curl_easy_perform (easyhandle);
+  CURLcode res (curl_easy_perform (easyhandle));
   curl_easy_cleanup (easyhandle);
   cout << readBuffer << endl;
 
   showTime ("Upload");
-
+  if (res != CURLE_OK)
+    cerr << prog << " failed: " << curl_easy_strerror (res) << endl;
   return 0;
 }
 
diff --git a/src/include/Attachment.hpp b/src/include/Attachment.hpp
index 0ffef82..62d0e20 100644
--- a/src/include/Attachment.hpp
+++ b/src/include/Attachment.hpp
@@ -35,9 +35,9 @@
 #ifndef _kaz_Attachment_hpp
 #define _kaz_Attachment_hpp
 
-#include <string>
-#include <regex>
 #include <map>
+#include <regex>
+#include <string>
 #include <utility>
 
 #include "EmbeddedData.hpp"