merged shared ut

src/utils/conftree.cpp

@@ -98,7 +98,11 @@ void ConfSimple::parseinput(istream& input)
 
         // Note that we trim whitespace before checking for backslash-eol
         // This avoids invisible whitespace problems.
-        trimstring(line);
+        if (trimvalues) {
+            trimstring(line);
+        } else {
+            ltrimstring(line);
+        }
         if (line.empty() || line.at(0) == '#') {
             if (eof) {
                 break;
@@ -119,7 +123,7 @@ void ConfSimple::parseinput(istream& input)
         appending = false;
 
         if (line[0] == '[') {
-            trimstring(line, "[]");
+            trimstring(line, "[] \t");
             if (dotildexpand) {
                 submapkey = path_tildexpand(line);
             } else {
@@ -142,7 +146,9 @@ void ConfSimple::parseinput(istream& input)
         nm = line.substr(0, eqpos);
         trimstring(nm);
         val = line.substr(eqpos + 1, string::npos);
-        trimstring(val);
+        if (trimvalues) {
+            trimstring(val);
+        }
 
         if (nm.length() == 0) {
             m_order.push_back(ConfLine(ConfLine::CFL_COMMENT, line));
@@ -156,8 +162,8 @@ void ConfSimple::parseinput(istream& input)
 }
 
 
-ConfSimple::ConfSimple(int readonly, bool tildexp)
+ConfSimple::ConfSimple(int readonly, bool tildexp, bool trimv)
-    : dotildexpand(tildexp), m_fmtime(0), m_holdWrites(false)
+    : dotildexpand(tildexp), trimvalues(trimv), m_fmtime(0), m_holdWrites(false)
 {
     status = readonly ? STATUS_RO : STATUS_RW;
 }
@@ -169,8 +175,8 @@ void ConfSimple::reparse(const string& d)
     parseinput(input);
 }
 
-ConfSimple::ConfSimple(const string& d, int readonly, bool tildexp)
+ConfSimple::ConfSimple(const string& d, int readonly, bool tildexp, bool trimv)
-    : dotildexpand(tildexp), m_fmtime(0), m_holdWrites(false)
+    : dotildexpand(tildexp), trimvalues(trimv), m_fmtime(0), m_holdWrites(false)
 {
     status = readonly ? STATUS_RO : STATUS_RW;
 
@@ -178,8 +184,10 @@ ConfSimple::ConfSimple(const string& d, int readonly, bool tildexp)
     parseinput(input);
 }
 
-ConfSimple::ConfSimple(const char *fname, int readonly, bool tildexp)
+ConfSimple::ConfSimple(const char *fname, int readonly, bool tildexp,
-    : dotildexpand(tildexp), m_filename(fname), m_fmtime(0), m_holdWrites(false)
+                       bool trimv)
+    : dotildexpand(tildexp), trimvalues(trimv), m_filename(fname),
+      m_fmtime(0), m_holdWrites(false)
 {
     status = readonly ? STATUS_RO : STATUS_RW;
 

src/utils/conftree.h

@@ -121,7 +121,8 @@ public:
      * @param readonly if true open readonly, else rw
      * @param tildexp  try tilde (home dir) expansion for subkey values
      */
-    ConfSimple(const char *fname, int readonly = 0, bool tildexp = false);
+    ConfSimple(const char *fname, int readonly = 0, bool tildexp = false,
+               bool trimvalues = true);
 
     /**
      * Build the object by reading content from a string
@@ -129,14 +130,16 @@ public:
      * @param readonly if true open readonly, else rw
      * @param tildexp  try tilde (home dir) expansion for subsection names
      */
-    ConfSimple(const string& data, int readonly = 0, bool tildexp = false);
+    ConfSimple(const string& data, int readonly = 0, bool tildexp = false,
+               bool trimvalues = true);
 
     /**
      * Build an empty object. This will be memory only, with no backing store.
      * @param readonly if true open read only, else rw
      * @param tildexp  try tilde (home dir) expansion for subsection names
      */
-    ConfSimple(int readonly = 0, bool tildexp = false);
+    ConfSimple(int readonly = 0, bool tildexp = false,
+               bool trimvalues = true);
 
     virtual ~ConfSimple() {};
 
@@ -281,6 +284,8 @@ public:
      */
     ConfSimple& operator=(const ConfSimple& rhs) {
         if (this != &rhs && (status = rhs.status) != STATUS_ERROR) {
+            dotildexpand = rhs.dotildexpand;
+            trimvalues = rhs.trimvalues;
             m_filename = rhs.m_filename;
             m_submaps = rhs.m_submaps;
         }
@@ -299,6 +304,7 @@ public:
     
 protected:
     bool dotildexpand;
+    bool trimvalues;
     StatusCode status;
 private:
     // Set if we're working with a file
@@ -345,12 +351,12 @@ class ConfTree : public ConfSimple {
 public:
     /* The constructors just call ConfSimple's, asking for key tilde
      * expansion */
-    ConfTree(const char *fname, int readonly = 0)
+    ConfTree(const char *fname, int readonly = 0, bool trimvalues=true)
-        : ConfSimple(fname, readonly, true) {}
+        : ConfSimple(fname, readonly, true, trimvalues) {}
-    ConfTree(const string& data, int readonly = 0)
+    ConfTree(const string& data, int readonly = 0, bool trimvalues=true)
-        : ConfSimple(data, readonly, true) {}
+        : ConfSimple(data, readonly, true, trimvalues) {}
-    ConfTree(int readonly = 0)
+    ConfTree(int readonly = 0, bool trimvalues=true)
-        : ConfSimple(readonly, true) {}
+        : ConfSimple(readonly, true, trimvalues) {}
     virtual ~ConfTree() {};
     ConfTree(const ConfTree& r) : ConfSimple(r) {};
     ConfTree& operator=(const ConfTree& r) {

src/utils/execmd.cpp

@@ -34,7 +34,9 @@
 #include <errno.h>
 #include <signal.h>
 #include <time.h>
+#include <string.h>
 
+#include <map>
 #include <vector>
 #include <string>
 #include <stdexcept>
@@ -493,27 +495,47 @@ int ExecCmd::startExec(const string& cmd, const vector<string>& args,
     }
     argv[i] = 0;
 
+    // Environment. We first merge our environment and the specified
+    // variables in a map<string,string>, overriding existing values,
+    // then generate an appropriate char*[]
     Ccharp *envv;
-    int envsize;
+    map<string, string> envmap;
-    for (envsize = 0; ; envsize++)
+    for (int i = 0; environ[i] != 0; i++) {
-        if (environ[envsize] == 0) {
+        string entry(environ[i]);
-            break;
+        string::size_type eqpos = entry.find_first_of("=");
+        if (eqpos == string::npos) {
+            continue;
         }
-    envv = (Ccharp *)malloc((envsize + m->m_env.size() + 2) * sizeof(char *));
+        envmap[entry.substr(0, eqpos)] = entry.substr(eqpos+1);
+    }
+    for (const auto& entry : m->m_env) {
+        string::size_type eqpos = entry.find_first_of("=");
+        if (eqpos == string::npos) {
+            continue;
+        }
+        envmap[entry.substr(0, eqpos)] = entry.substr(eqpos+1);
+    }        
+
+    // Allocate space for the array + string storage in one block.
+    unsigned int allocsize = (envmap.size() + 2) * sizeof(char *);
+    for (const auto& it : envmap) {
+        allocsize += it.first.size() + 1 + it.second.size() + 1;
+    }
+    envv = (Ccharp *)malloc(allocsize);
     if (envv == 0) {
         LOGERR("ExecCmd::doexec: malloc() failed. errno " << errno << "\n");
         free(argv);
         return -1;
     }
-    int eidx;
+    // Copy to new env array
-    for (eidx = 0; eidx < envsize; eidx++) {
+    i = 0;
-        envv[eidx] = environ[eidx];
+    char *cp = ((char *)envv) + (envmap.size() + 2) * sizeof(char *);
+    for (const auto& it : envmap) {
+        strcpy(cp, (it.first + "=" + it.second).c_str());
+        envv[i++] = cp;
+        cp += it.first.size() + 1 + it.second.size() + 1;
     }
-    for (vector<string>::const_iterator it = m->m_env.begin();
+    envv[i++] = 0;
-            it != m->m_env.end(); it++) {
-        envv[eidx++] = it->c_str();
-    }
-    envv[eidx] = 0;
 
     // As we are going to use execve, not execvp, do the PATH thing.
     string exe;

src/utils/pathut.cpp

@@ -15,7 +15,6 @@
  *   59 Temple Place - Suite 330, Boston, MA  02111-1307, USA.
  */
 
-#ifndef TEST_PATHUT
 #ifdef BUILDING_RECOLL
 #include "autoconfig.h"
 #else
@@ -612,6 +611,45 @@ string url_encode(const string& url, string::size_type offs)
     return out;
 }
 
+static inline int h2d(int c) {
+    if ('0' <= c && c <= '9')
+        return c - '0';
+    else if ('A' <= c && c <= 'F')
+        return 10 + c - 'A';
+    else 
+        return -1;
+}
+
+string url_decode(const string &in)
+{
+    if (in.size() <= 2)
+        return in;
+    string out;
+    out.reserve(in.size());
+    const char *cp = in.c_str();
+    string::size_type i = 0;
+    for (; i < in.size() - 2; i++) {
+	if (cp[i] == '%') {
+            int d1 = h2d(cp[i+1]);
+            int d2 = h2d(cp[i+2]);
+            if (d1 != -1 && d2 != -1) {
+                out += (d1 << 4) + d2;
+            } else {
+                out += '%';
+                out += cp[i+1];
+                out += cp[i+2];
+            }
+            i += 2;
+	} else {
+            out += cp[i];
+        }
+    }
+    while (i < in.size()) {
+        out += cp[i++];
+    }
+    return out;
+}
+
 string url_gpath(const string& url)
 {
     // Remove the access schema part (or whatever it's called)
@@ -867,198 +905,3 @@ void pathut_init_mt()
 {
     path_home();
 }
-
-
-#else // TEST_PATHUT
-#include <stdlib.h>
-#include <iostream>
-using namespace std;
-
-#include "pathut.h"
-
-void path_to_thumb(const string& _input)
-{
-    string input(_input);
-    // Make absolute path if needed
-    if (input[0] != '/') {
-        input = path_absolute(input);
-    }
-
-    input = string("file://") + path_canon(input);
-
-    string path;
-    //path = url_encode(input, 7);
-    thumbPathForUrl(input, 7, path);
-    cout << path << endl;
-}
-
-const char *tstvec[] = {"", "/", "/dir", "/dir/", "/dir1/dir2",
-                        "/dir1/dir2",
-                        "./dir", "./dir1/", "dir", "../dir", "/dir/toto.c",
-                        "/dir/.c", "/dir/toto.txt", "toto.txt1"
-                       };
-
-const string ttvec[] = {"/dir", "", "~", "~/sub", "~root", "~root/sub",
-                        "~nosuch", "~nosuch/sub"
-                       };
-int nttvec = sizeof(ttvec) / sizeof(string);
-
-const char *thisprog;
-
-int main(int argc, const char **argv)
-{
-    thisprog = *argv++;
-    argc--;
-
-    string s;
-    vector<string>::const_iterator it;
-#if 0
-    for (unsigned int i = 0; i < sizeof(tstvec) / sizeof(char *); i++) {
-        cout << tstvec[i] << " Father " << path_getfather(tstvec[i]) << endl;
-    }
-    for (unsigned int i = 0; i < sizeof(tstvec) / sizeof(char *); i++) {
-        cout << tstvec[i] << " Simple " << path_getsimple(tstvec[i]) << endl;
-    }
-    for (unsigned int i = 0; i < sizeof(tstvec) / sizeof(char *); i++) {
-        cout << tstvec[i] << " Basename " <<
-             path_basename(tstvec[i], ".txt") << endl;
-    }
-#endif
-
-#if 0
-    for (int i = 0; i < nttvec; i++) {
-        cout << "tildexp: '" << ttvec[i] << "' -> '" <<
-             path_tildexpand(ttvec[i]) << "'" << endl;
-    }
-#endif
-
-#if 0
-    const string canontst[] = {"/dir1/../../..", "/////", "",
-                               "/dir1/../../.././/////dir2///////",
-                               "../../",
-                               "../../../../../../../../../../"
-                              };
-    unsigned int nttvec = sizeof(canontst) / sizeof(string);
-    for (unsigned int i = 0; i < nttvec; i++) {
-        cout << "canon: '" << canontst[i] << "' -> '" <<
-             path_canon(canontst[i]) << "'" << endl;
-    }
-#endif
-#if 0
-    if (argc != 2) {
-        cerr << "Usage: trpathut <dir> <pattern>" << endl;
-        exit(1);
-    }
-    string dir = *argv++;
-    argc--;
-    string pattern =  *argv++;
-    argc--;
-    vector<string> matched = path_dirglob(dir, pattern);
-    for (it = matched.begin(); it != matched.end(); it++) {
-        cout << *it << endl;
-    }
-#endif
-
-#if 0
-    if (argc != 1) {
-        fprintf(stderr, "Usage: fsocc: trpathut <path>\n");
-        exit(1);
-    }
-    string path = *argv++;
-    argc--;
-
-    int pc;
-    long long blocks;
-    if (!fsocc(path, &pc, &blocks)) {
-        fprintf(stderr, "fsocc failed\n");
-        return 1;
-    }
-    printf("pc %d, megabytes %ld\n", pc, blocks);
-#endif
-
-#if 0
-    Pidfile pidfile("/tmp/pathutpidfile");
-    pid_t pid;
-    if ((pid = pidfile.open()) != 0) {
-        cerr << "open failed. reason: " << pidfile.getreason() <<
-             " return " << pid << endl;
-        exit(1);
-    }
-    pidfile.write_pid();
-    sleep(10);
-    pidfile.close();
-    pidfile.remove();
-#endif
-
-#if 0
-    if (argc > 1) {
-        cerr <<  "Usage: thumbpath <filepath>" << endl;
-        exit(1);
-    }
-    string input;
-    if (argc == 1) {
-        input = *argv++;
-        if (input.empty())  {
-            cerr << "Usage: thumbpath <filepath>" << endl;
-            exit(1);
-        }
-        path_to_thumb(input);
-    } else {
-        while (getline(cin, input)) {
-            path_to_thumb(input);
-        }
-    }
-
-
-    exit(0);
-#endif
-
-#if 0
-    if (argc != 1) {
-        cerr << "Usage: trpathut <filename>" << endl;
-        exit(1);
-    }
-    string fn = *argv++;
-    argc--;
-    string ext = path_suffix(fn);
-    cout << "Suffix: [" << ext << "]" << endl;
-    return 0;
-#endif
-
-#if 0
-    if (argc != 1) {
-        cerr << "Usage: trpathut url" << endl;
-        exit(1);
-    }
-    string url = *argv++;
-    argc--;
-
-    cout << "File: [" << fileurltolocalpath(url) << "]\n";
-    return 0;
-#endif
-
-#if 1
-    if (argc != 1) {
-        cerr << "Usage: trpathut path" << endl;
-        exit(1);
-    }
-    string path = *argv++;
-    argc--;
-
-    int pc;
-    long long avmbs;
-    if (fsocc(path, &pc, &avmbs)) {
-        cout << "Percent occup " << pc << " avmbs " << avmbs << endl;
-        return 0;
-    } else {
-        cerr << "fsocc failed\n";
-        return 1;
-    }
-#endif
-
-
-
-}
-
-#endif // TEST_PATHUT
-

src/utils/pathut.h

@@ -52,6 +52,7 @@ extern std::vector<std::string> path_dirglob(const std::string& dir,
 /// Encode according to rfc 1738
 extern std::string url_encode(const std::string& url,
                               std::string::size_type offs = 0);
+extern std::string url_decode(const std::string& encoded);
 //// Convert to file path if url is like file://. This modifies the
 //// input (and returns a copy for convenience)
 extern std::string fileurltolocalpath(std::string url);

src/utils/smallut.cpp

@@ -453,6 +453,20 @@ bool stringToBool(const string& s)
 }
 
 void trimstring(string& s, const char *ws)
+{
+    rtrimstring(s, ws);
+    ltrimstring(s, ws);
+}
+
+void rtrimstring(string& s, const char *ws)
+{
+    string::size_type pos = s.find_last_not_of(ws);
+    if (pos != string::npos && pos != s.length() - 1) {
+        s.replace(pos + 1, string::npos, string());
+    }
+}
+
+void ltrimstring(string& s, const char *ws)
 {
     string::size_type pos = s.find_first_not_of(ws);
     if (pos == string::npos) {
@@ -460,11 +474,6 @@ void trimstring(string& s, const char *ws)
         return;
     }
     s.replace(0, pos, string());
-
-    pos = s.find_last_not_of(ws);
-    if (pos != string::npos && pos != s.length() - 1) {
-        s.replace(pos + 1, string::npos, string());
-    }
 }
 
 // Remove some chars and replace them with spaces

src/utils/smallut.h

@@ -142,6 +142,8 @@ extern bool stringToBool(const std::string& s);
 /** Remove instances of characters belonging to set (default {space,
     tab}) at beginning and end of input string */
 extern void trimstring(std::string& s, const char *ws = " \t");
+extern void rtrimstring(std::string& s, const char *ws = " \t");
+extern void ltrimstring(std::string& s, const char *ws = " \t");
 
 /** Escape things like < or & by turning them into entities */
 extern std::string escapeHtml(const std::string& in);