/*
 * usage: ./dependencyDiscoverer [-Idir] ... file.c|file.l|file.y ...
 *
 * processes the c/yacc/lex source file arguments, outputting the dependencies
 * between the corresponding .o file, the .c source file, and any included
 * .h files
 *
 * each .h file is also processed to yield a dependency between it and any
 * included .h files
 *
 * these dependencies are written to standard output in a form compatible with
 * make; for example, assume that foo.c includes inc1.h, and inc1.h includes
 * inc2.h and inc3.h; this results in
 *
 *                  foo.o: foo.c inc1.h inc2.h inc3.h
 *
 * note that system includes (i.e. those in angle brackets) are NOT processed
 *
 * dependencyDiscoverer uses the CPATH environment variable, which can contain a
 * set of directories separated by ':' to find included files
 * if any additional directories are specified in the command line,
 * these are prepended to those in CPATH, left to right
 *
 * for example, if CPATH is "/home/user/include:/usr/local/group/include",
 * and if "-Ifoo/bar/include" is specified on the command line, then when
 * processing
 *           #include "x.h"
 * x.h will be located by searching for the following files in this order
 *
 *      ./x.h
 *      foo/bar/include/x.h
 *      /home/user/include/x.h
 *      /usr/local/group/include/x.h
 */

/*
 * general design of main()
 * ========================
 * There are three globally accessible variables:
 * - dirs: a vector storing the directories to search for headers
 * - theTable: a hash table mapping file names to a list of dependent file names
 * - workQ: a list of file names that have to be processed
 *
 * 1. look up CPATH in environment
 * 2. assemble dirs vector from ".", any -Idir flags, and fields in CPATH
 *    (if it is defined)
 * 3. for each file argument (after -Idir flags)
 *    a. insert mapping from file.o to file.ext (where ext is c, y, or l) into
 *       table
 *    b. insert mapping from file.ext to empty list into table
 *    c. append file.ext on workQ
 * 4. for each file on the workQ
 *    a. lookup list of dependencies
 *    b. invoke process(name, list_of_dependencies)
 * 5. for each file argument (after -Idir flags)
 *    a. create a hash table in which to track file names already printed
 *    b. create a linked list to track dependencies yet to print
 *    c. print "foo.o:", insert "foo.o" into hash table
 *       and append "foo.o" to linked list
 *    d. invoke printDependencies()
 *
 * general design for process()
 * ============================
 *
 * 1. open the file
 * 2. for each line of the file
 *    a. skip leading whitespace
 *    b. if match "#include"
 *       i. skip leading whitespace
 *       ii. if next character is '"'
 *           * collect remaining characters of file name (up to '"')
 *           * append file name to dependency list for this open file
 *           * if file name not already in the master Table
 *             - insert mapping from file name to empty list in master table
 *             - append file name to workQ
 * 3. close file
 *
 * general design for printDependencies()
 * ======================================
 *
 * 1. while there is still a file in the toProcess linked list
 * 2. fetch next file from toProcess
 * 3. lookup up the file in the master table, yielding the linked list of dependencies
 * 4. iterate over dependenceies
 *    a. if the filename is already in the printed hash table, continue
 *    b. print the filename
 *    c. insert into printed
 *    d. append to toProcess
 *
 * Additional helper functions
 * ===========================
 *
 * dirName() - appends trailing '/' if needed
 * parseFile() - breaks up filename into root and extension
 * openFile()  - attempts to open a filename using the search path defined by the dirs vector.
 */

#include <ctype.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>

#include <string>
#include <vector>
#include <unordered_map>
#include <unordered_set>
#include <list>
#include <mutex>
#include <optional>
#include <condition_variable>
#include <thread>
#include <experimental/optional>

std::vector<std::string> dirs;
hashTable theTable;
work_q workQ;


struct hashTable{
private:
    std::unordered_map<std::string,std::list<std::string>> map;
    std::mutex mutex;
    std::condition_variable ready;
public:
    void insert(std::string key,std::list<std::string> value) {
        {
            std::unique_lock <std::mutex> lock(mutex);
            map.insert({key, {value}});
        }
        ready.notify_one();
    }

    std::list<std::string> *at(std::string key){
        {
            std::unique_lock<std::mutex> lock(mutex);
            return(&map[key]);
        }
    }
    bool check(std::string key)
    {
        {
            std::unique_lock<std::mutex> lock(mutex);
            if(map.find(key) != map.end()){
                return (true);
            }
            return (false);
        }
    }
};

struct work_q {
private:
    std::list<std::string> q;
    std::mutex m;
    std::condition_variable ready;
    bool done = false;
public:
  std::optional<std::string> pop(){
    std::unique_lock<std::mutex> lock(m);
    ready.wait(lock, [this]{ return (!q.empty()) || done; });
    if (q.empty()) return {};
    auto f = q.front();
    q.pop_front();
    return f;
}
void push(std::string f) {
    {
        std::unique_lock<std::mutex> lock(m);
        workQ.push(f);
    }
    ready.notify_one();
}
void setDone() {
    {
        std::unique_lock<std::mutex> lock(m);
        done = true;
    }
    ready.notify_all();
}

int getSize()
{
  std::unique_lock<std::mutex> lock(m);
  int size = q.size();
  return(size);
}
};



struct task_system {
private:
    int count;
    std::vector<std::thread> threads;
    void run(int i) {
        while (true) {
            auto optional_f = theTable.pop();
            if (!optional_f.has_value()) return;
            auto filename = optional_f.value();
            process(filename.c_str(), theTable.look(optional_f));
        }
    }
public:
    task_system(int c){
        count = c;
        //printf("Start task system with %d threads\n", count);
        for (auto n = 0; n != count; n++) {
            threads.emplace_back([this, n](){ run(n); }); }
    }
    ~task_system() {
        workQ.setDone();
        for (auto n = 0; n != count; n++) { threads[n].join(); }
    }
    void async(std::string f) {
        workQ.push(f);
    }
};



std::string dirName(const char * c_str) {
  std::string s = c_str; // s takes ownership of the string content by allocating memory for it
  if (s.back() != '/') { s += '/'; }
  return s;
}

std::pair<std::string, std::string> parseFile(const char* c_file) {
  std::string file = c_file;
  std::string::size_type pos = file.rfind('.');
  if (pos == std::string::npos) {
    return {file, ""};
  } else {
    return {file.substr(0, pos), file.substr(pos + 1)};
  }
}

// open file using the directory search path constructed in main()
static FILE *openFile(const char *file) {
  FILE *fd;
  for (unsigned int i = 0; i < dirs.size(); i++) {
    std::string path = dirs[i] + file;
    fd = fopen(path.c_str(), "r");
    if (fd != NULL)
      return fd; // return the first file that successfully opens
  }
  return NULL;
}

// process file, looking for #include "foo.h" lines
static void process(const char *file, std::list<std::string> *ll) {
  char buf[4096], name[4096];
  // 1. open the file
  FILE *fd = openFile(file);
  if (fd == NULL) {
    fprintf(stderr, "Error opening %s\n", file);
    exit(-1);
  }
  while (fgets(buf, sizeof(buf), fd) != NULL) {
    char *p = buf;
    // 2a. skip leading whitespace
    while (isspace((int)*p)) { p++; }
    // 2b. if match #include
    if (strncmp(p, "#include", 8) != 0) { continue; }
    p += 8; // point to first character past #include
    // 2bi. skip leading whitespace
    while (isspace((int)*p)) { p++; }
    if (*p != '"') { continue; }
    // 2bii. next character is a "
    p++; // skip "
    // 2bii. collect remaining characters of file name
    char *q = name;
    while (*p != '\0') {
      if (*p == '"') { break; }
      *q++ = *p++;
    }
    *q = '\0';
    // 2bii. append file name to dependency list
    ll->push_back( {name} );
    // 2bii. if file name not already in table ...
    if (theTable.check(name) != theTable.end()) { continue; }
    // ... insert mapping from file name to empty list in table ...
    workQ.insert( { name, {} } );
    // ... append file name to workQ
    workQ.push( name );
  }
  // 3. close file
  fclose(fd);
}

// iteratively print dependencies
static void printDependencies(std::unordered_set<std::string> *printed,
                              std::list<std::string> *toProcess,
                              FILE *fd) {
  if (!printed || !toProcess || !fd) return;

  // 1. while there is still a file in the toProcess list
  while ( toProcess->size() > 0 ) {
    // 2. fetch next file to process
    std::string name = toProcess->front();
    toProcess->pop_front();
    // 3. lookup file in the table, yielding list of dependencies
    std::list<std::string> *ll = &theTable[name];
    // 4. iterate over dependencies
    for (auto iter = ll->begin(); iter != ll->end(); iter++) {
      // 4a. if filename is already in the printed table, continue
      if (printed->find(*iter) != printed->end()) { continue; }
      // 4b. print filename
      fprintf(fd, " %s", iter->c_str());
      // 4c. insert into printed
      printed->insert( *iter );
      // 4d. append to toProcess
      toProcess->push_back( *iter );
    }
  }
}

int main(int argc, char *argv[]) {
    // 1. look up CPATH in environment
    char *cpath = getenv("CPATH");
    int count =2;
    // determine the number of -Idir arguments
    int i;
    for (i = 1; i < argc; i++) {
        if (strncmp(argv[i], "-I", 2) != 0)
            break;
    }
    int start = i;

    // 2. start assembling dirs vector
    dirs.push_back(dirName("./")); // always search current directory first
    for (i = 1; i < start; i++) {
        dirs.push_back(dirName(argv[i] + 2 /* skip -I */));
    }
    if (cpath != NULL) {
        std::string str(cpath);
        std::string::size_type last = 0;
        std::string::size_type next = 0;
        while ((next = str.find(":", last)) != std::string::npos) {
            dirs.push_back(str.substr(last, next - last));
            last = next + 1;
        }
        dirs.push_back(str.substr(last));
    }
    // 2. finished assembling dirs vector
    task_system tasks(count);
    // 3. for each file argument ...
    for (i = start; i < argc; i++) {
        std::pair <std::string, std::string> pair = parseFile(argv[i]);
        if (pair.second != "c" && pair.second != "y" && pair.second != "l") {
            fprintf(stderr, "Illegal extension: %s - must be .c, .y or .l\n",
                    pair.second.c_str());
            return -1;
        }

        std::string obj = pair.first + ".o";

        // 3a. insert mapping from file.o to file.ext
        theTable.insert({obj, {argv[i]}});

        // 3b. insert mapping from file.ext to empty list
        theTable.insert({argv[i], {}});

        // 3c. append file.ext on workQ
        workQ.push(argv[i]);
    }


  // 4. for each file on the workQ
  while ( workQ.size() > 0 ) {
    std::string filename = workQ.front();
    workQ.pop();

    if (theTable.find(filename) == theTable.end()) {
      fprintf(stderr, "Mismatch between table and workQ\n");
      return -1;
    }

    // 4a&b. lookup dependencies and invoke 'process'
    process(filename.c_str(), &theTable[filename]);
  }

  // 5. for each file argument
  for (i = start; i < argc; i++) {
    // 5a. create hash table in which to track file names already printed
    std::unordered_set<std::string> printed;
    // 5b. create list to track dependencies yet to print
    std::list<std::string> toProcess;

    std::pair<std::string, std::string> pair = parseFile(argv[i]);

    std::string obj = pair.first + ".o";
    // 5c. print "foo.o:" ...
    printf("%s:", obj.c_str());
    // 5c. ... insert "foo.o" into hash table and append to list
    printed.insert( obj );
    toProcess.push_back( obj );
    // 5d. invoke
    printDependencies(&printed, &toProcess, stdout);

    printf("\n");
  }

  return 0;
}