indexing: allow setting a maxseconds parameter on mimeconf lines. Allows setting a specific 30S value for text/man which should be fast but sometimes goes looping

src/internfile/mh_exec.cpp

@@ -63,9 +63,7 @@ void MEAdv::newData(int n)
 
 
 MimeHandlerExec::MimeHandlerExec(RclConfig *cnf, const std::string& id)
-    : RecollFilter(cnf, id), missingHelper(false), m_filtermaxseconds(900),
+    : RecollFilter(cnf, id)
-      m_filtermaxmbytes(0), m_handlernomd5(false), m_hnomd5init(false),
-      m_nomd5(false)
 {
     m_config->getConfParam("filtermaxseconds", &m_filtermaxseconds);
     m_config->getConfParam("filtermaxmbytes", &m_filtermaxmbytes);

src/internfile/mh_exec.h

@@ -44,7 +44,7 @@ class HandlerTimeout {};
  * (cmd, params etc.)
  */
 class MimeHandlerExec : public RecollFilter {
- public:
+public:
     ///////////////////////
     // Members not reset by clear(). params, cfgFilterOutputMtype and 
     // cfgFilterOutputCharset
@@ -62,33 +62,42 @@ class MimeHandlerExec : public RecollFilter {
     // those filters, the output charset has to be known: ie set by a command
     // line option.
     std::string cfgFilterOutputCharset; 
-    bool missingHelper;
+    bool missingHelper{false};
     // Resource management values
-    int m_filtermaxseconds;
+
-    int m_filtermaxmbytes;
+    // The filtermaxseconds default is set in the constructor by
+    // querying the recoll.conf configuration variable. It can be
+    // changed by the filter creation code in mimehandler.cpp if a
+    // maxseconds parameter is set on the mimeconf line.
+    int m_filtermaxseconds{900};
+    int m_filtermaxmbytes{0};
     ////////////////
 
     MimeHandlerExec(RclConfig *cnf, const std::string& id);
 
+    virtual void setmaxseconds(int seconds) {
+        m_filtermaxseconds = seconds;
+    }
+    
     virtual bool next_document() override;
     virtual bool skip_to_document(const std::string& ipath) override;
 
     virtual void clear_impl() override {
-    m_fn.erase(); 
+        m_fn.erase(); 
-    m_ipath.erase();
+        m_ipath.erase();
     }
 
 protected:
-    virtual bool set_document_file_impl(const std::string& mt, 
+    virtual bool set_document_file_impl(
-                                        const std::string& file_path) override;
+        const std::string& mt, const std::string& file_path) override;
 
     std::string m_fn;
     std::string m_ipath;
     // md5 computation excluded by handler name: can't change after init
-    bool m_handlernomd5;
+    bool m_handlernomd5{false};
-    bool m_hnomd5init;
+    bool m_hnomd5init{false};
     // If md5 not excluded by handler name, allow/forbid depending on mime
-    bool m_nomd5;
+    bool m_nomd5{false};
     
     // Set the character set field and possibly transcode text/plain
     // output.

src/internfile/mimehandler.cpp

@@ -202,6 +202,7 @@ static RecollFilter *mhFactory(RclConfig *config, const string &mimeOrParams,
 }
 
 static const string cstr_mh_charset("charset");
+static const string cstr_mh_maxseconds("maxseconds");
 /**
  * Create a filter that executes an external program or script
  * A filter def can look like:
@@ -245,7 +246,9 @@ MimeHandlerExec *mhExecFactory(RclConfig *cfg, const string& mtype, string& hs,
         h->cfgFilterOutputCharset = stringtolower((const string&)value);
     if (attrs.get(cstr_dj_keymt, value))
         h->cfgFilterOutputMtype = stringtolower((const string&)value);
-
+    if (attrs.get(cstr_mh_maxseconds, value)) {
+        h->setmaxseconds(atoi(value.c_str()));
+    }
     LOGDEB2("mhExecFactory:mt [" << mtype << "] cfgmt [" <<
             h->cfgFilterOutputMtype << "] cfgcs ["<<h->cfgFilterOutputCharset <<
             "] cmd: [" << stringsToString(h->params) << "]\n");

src/sampleconf/mimeconf

@@ -222,7 +222,7 @@ text/x-ini = internal text/plain
 text/x-java = internal text/plain
 text/x-lua = internal
 text/x-mail = internal
-text/x-man = exec rclman
+text/x-man = exec rclman;maxseconds=30
 text/x-orgmode = execm rclorgmode.py
 text/x-perl = internal text/plain
 text/x-purple-html-log = internal text/html