unify processing for executing a python script

src/common/rclconfig.cpp

@@ -1677,6 +1677,43 @@ string RclConfig::findFilter(const string &icmd) const
     }
 }
 
+bool RclConfig::processFilterCmd(std::vector<std::string>& cmd) const
+{
+    LOGDEB0("processFilterCmd: in: " << stringsToString(cmd) << "\n");
+    auto it = cmd.begin();
+
+    // Special-case python and perl on windows: we need to also locate the
+    // first argument which is the script name "python somescript.py". 
+    // On Unix, thanks to #!, we usually just run "somescript.py", but need
+    // the same change if we ever want to use the same cmd line as windows
+    bool hasinterp = !stringlowercmp("python", *it) ||
+        !stringlowercmp("perl", *it);
+
+    *it++ = findFilter(*it);
+    if (hasinterp) {
+        if (cmd.size() < 2) {
+            LOGERR("processFilterCmd: python/perl cmd: no script?. [" <<
+                   stringsToString(cmd) << "]\n");
+            return false;
+        } else {
+            *it = findFilter(*it);
+        }
+    }
+    LOGDEB0("processFilterCmd: out: " << stringsToString(cmd) << "\n");
+    return true;
+}
+
+bool RclConfig::pythonCmd(const std::string& scriptname,
+                          std::vector<std::string>& cmd) const
+{
+#ifdef _WIN32
+    cmd = {"python", scriptname};
+#else
+    cmd = {scriptname};
+#endif
+    return processFilterCmd(cmd);
+}
+
 /** 
  * Return decompression command line for given mime type
  */
@@ -1693,32 +1730,14 @@ bool RclConfig::getUncompressor(const string &mtype, vector<string>& cmd) const
         LOGERR("getUncompressor: empty spec for mtype " << mtype << "\n");
         return false;
     }
-    vector<string>::iterator it = tokens.begin();
+    auto it = tokens.begin();
     if (tokens.size() < 2)
         return false;
     if (stringlowercmp("uncompress", *it++)) 
         return false;
     cmd.clear();
-    cmd.push_back(findFilter(*it));
-
-    // Special-case python and perl on windows: we need to also locate the
-    // first argument which is the script name "python somescript.py". 
-    // On Unix, thanks to #!, we usually just run "somescript.py", but need
-    // the same change if we ever want to use the same cmdling as windows
-    if (!stringlowercmp("python", *it) || !stringlowercmp("perl", *it)) {
-        it++;
-        if (tokens.size() < 3) {
-            LOGERR("getUncpressor: python/perl cmd: no script?. [" <<
-                   mtype << "]\n");
-        } else {
-            *it = findFilter(*it);
-        }
-    } else {
-        it++;
-    }
-    
     cmd.insert(cmd.end(), it, tokens.end());
-    return true;
+    return processFilterCmd(cmd);
 }
 
 static const char blurb0[] = 

src/common/rclconfig.h

@@ -332,6 +332,16 @@ class RclConfig {
     bool getMissingHelperDesc(string&) const;
     void storeMissingHelperDesc(const string &s);
 
+    /** Replace simple command name(s) inside vector with full
+     * paths. May have to replace two if the first entry is an
+     * interpreter name */
+    bool processFilterCmd(std::vector<std::string>& cmd) const;
+
+    /** Build command vector for python script, possibly prepending 
+        interpreter on Windows */
+    bool pythonCmd(
+        const std::string& script,  std::vector<std::string>& cmd) const;
+    
     /** Find exec file for external filter. 
      *
      * If the input is an absolute path, we just return it. Else We

src/common/textsplitko.cpp

@@ -59,13 +59,13 @@ static const string magicpage{"NEWPPPAGE"};
 
 void TextSplit::koStaticConfInit(RclConfig *config, const string& tagger)
 {
-#ifdef _WIN32
-    o_cmdpath = config->findFilter("python");
-    o_cmdargs.clear();
-    o_cmdargs.push_back(config->findFilter("kosplitter.py"));
-#else
-    o_cmdpath = config->findFilter("kosplitter.py");
-#endif
+    std::vector<std::string> cmdvec;
+    if (config->pythonCmd("kosplitter.py", cmdvec)) {
+        auto it = cmdvec.begin();
+        o_cmdpath = *it++;
+        o_cmdargs.clear();
+        o_cmdargs.insert(o_cmdargs.end(), it, cmdvec.end());
+    }
     if (tagger == "Okt" || tagger == "Mecab" || tagger == "Komoran") {
         o_taggername = tagger;
         if (tagger == "Komoran")

src/index/indexer.cpp

@@ -66,22 +66,9 @@ bool runWebFilesMoverScript(RclConfig *config)
             downloadsdir = path_tildexpand("~/Downloads");
         }
     }
-    static string cmdpath;
-    vector<string> args;
-#ifdef _WIN32
-    const static string cmdnm{"python"};
-    args.push_back(config->findFilter("recoll-we-move-files.py"));
-#else
-    const static string cmdnm{"recoll-we-move-files.py"};
-#endif
-    if (cmdpath.empty()) {
-        cmdpath = config->findFilter(cmdnm);
-        if (cmdpath.empty()) {
-            LOGERR("runWFMoverScript: recoll-we-move-files.py not found\n");
-            return false;
-        }
-    }
-
+    vector<string> cmdvec;
+    config->pythonCmd("recoll-we-move-files.py", cmdvec);
+    
     /* Arrange to not actually run the script if the directory did not change */
     static time_t dirmtime;
     time_t ndirmtime = 0;
@@ -100,7 +87,7 @@ bool runWebFilesMoverScript(RclConfig *config)
            are created during the run. */
         dirmtime = ndirmtime;
         ExecCmd cmd;
-        int status = cmd.doexec(cmdpath, args);
+        int status = cmd.doexec1(cmdvec);
         return status == 0;
     }
     return true;

src/internfile/mimehandler.cpp

@@ -231,27 +231,12 @@ MimeHandlerExec *mhExecFactory(RclConfig *cfg, const string& mtype, string& hs,
                "]: [" << hs << "]\n");
         return 0;
     }
-    MimeHandlerExec *h = multiple ? 
-        new MimeHandlerExecMultiple(cfg, id) :
-        new MimeHandlerExec(cfg, id);
-    vector<string>::iterator it = cmdtoks.begin();
-
-    // Special-case python and perl on windows: we need to also locate the
-    // first argument which is the script name "python somescript.py". 
-    // On Unix, thanks to #!, we usually just run "somescript.py", but need
-    // the same change if we ever want to use the same cmdling as windows
-    if (!stringlowercmp("python", *it) || !stringlowercmp("perl", *it)) {
-        if (cmdtoks.size() < 2) {
-            LOGERR("mhExecFactory: python/perl cmd: no script?. [" <<
-                   mtype << "]: [" << hs << "]\n");
-        }
-        vector<string>::iterator it1(it);
-        it1++;
-        *it1 = cfg->findFilter(*it1);
+    if (!cfg->processFilterCmd(cmdtoks)) {
+        return nullptr;
     }
-            
-    h->params.push_back(cfg->findFilter(*it++));
-    h->params.insert(h->params.end(), it, cmdtoks.end());
+    MimeHandlerExec *h = multiple ? new MimeHandlerExecMultiple(cfg, id) :
+        new MimeHandlerExec(cfg, id);
+    h->params = cmdtoks;
 
     // Handle additional attributes. We substitute the semi-colons
     // with newlines and use a ConfSimple
@@ -261,16 +246,9 @@ MimeHandlerExec *mhExecFactory(RclConfig *cfg, const string& mtype, string& hs,
     if (attrs.get(cstr_dj_keymt, value))
         h->cfgFilterOutputMtype = stringtolower((const string&)value);
 
-#if 0
-    string scmd;
-    for (it = h->params.begin(); it != h->params.end(); it++) {
-        scmd += string("[") + *it + "] ";
-    }
-    LOGDEB("mhExecFactory:mt [" << mtype << "] cfgmt [" <<
-           h->cfgFilterOutputMtype << "] cfgcs [" <<
-           h->cfgFilterOutputCharset << "] cmd: [" << scmd << "]\n");
-#endif
-
+    LOGDEB2("mhExecFactory:mt [" << mtype << "] cfgmt [" <<
+            h->cfgFilterOutputMtype << "] cfgcs ["<<h->cfgFilterOutputCharset <<
+            "] cmd: [" << stringsToString(h->params) << "]\n");
     return h;
 }