about summary refs log tree commit diff stats
diff options
context:
space:
mode:
Diffstat
-rw-r--r--Makefile.am4
-rw-r--r--config-example.yml6
-rw-r--r--configure.ac28
-rw-r--r--kgramstats.cpp110
-rw-r--r--kgramstats.h28
-rw-r--r--main.cpp161
6 files changed, 336 insertions, 1 deletions
diff --git a/Makefile.am b/Makefile.am index 127042a..c5b52ce 100644 --- a/Makefile.am +++ b/Makefile.am
@@ -2,4 +2,6 @@ AUTOMAKE_OPTIONS = subdir-objects
2ACLOCAL_AMFLAGS = ${ACLOCAL_FLAGS} 2ACLOCAL_AMFLAGS = ${ACLOCAL_FLAGS}
3 3
4bin_PROGRAMS = rawr-ebooks 4bin_PROGRAMS = rawr-ebooks
5rawr_ebooks_SOURCES = main.cpp \ No newline at end of file 5rawr_ebooks_SOURCES = main.cpp kgramstats.cpp
6AM_CPPFLAGS = $(LIBTWITCURL_CFLAGS) $(YAML_CFLAGS)
7rawr_ebooks_LDADD = $(LIBTWITCURL_LIBS) $(YAML_LIBS) \ No newline at end of file
diff --git a/config-example.yml b/config-example.yml new file mode 100644 index 0000000..a280e4c --- /dev/null +++ b/config-example.yml
@@ -0,0 +1,6 @@
1---
2 corpus: "corpus.txt"
3 username: ""
4 password: ""
5 consumer_key: ""
6 consumer_secret: "" \ No newline at end of file
diff --git a/configure.ac b/configure.ac index c1c50bf..31cf8b9 100644 --- a/configure.ac +++ b/configure.ac
@@ -4,4 +4,32 @@ AM_INIT_AUTOMAKE([1.10 no-define foreign])
4AC_PROG_CXX 4AC_PROG_CXX
5AC_CONFIG_FILES([Makefile]) 5AC_CONFIG_FILES([Makefile])
6 6
7# Get libtwitcurl library and include locations
8AC_ARG_WITH([libtwitcurl-include-path],
9 [AS_HELP_STRING([--with-libtwitcurl-include-path],
10 [location of the libtwitcurl headers, defaults to /usr/include])],
11 [LIBTWITCURL_CFLAGS="-I$withval"],
12 [LIBTWITCURL_CFLAGS=''])
13AC_SUBST([LIBTWITCURL_CFLAGS])
14
15AC_ARG_WITH([libtwitcurl-lib-path],
16 [AS_HELP_STRING([--with-libtwitcurl-lib-path], [location of the libtwitcurl libraries])],
17 [LIBTWITCURL_LIBS="-L$withval -llibtwitcurl"],
18 [LIBTWITCURL_LIBS='-ltwitcurl'])
19AC_SUBST([LIBTWITCURL_LIBS])
20
21# Get yaml-cpp library and include locations
22AC_ARG_WITH([yamlcpp-include-path],
23 [AS_HELP_STRING([--with-yamlcpp-include-path],
24 [location of the yamlcpp headers, defaults to /usr/include])],
25 [YAML_CFLAGS="-I$withval"],
26 [YAML_CFLAGS=''])
27AC_SUBST([YAML_CFLAGS])
28
29AC_ARG_WITH([yamlcpp-lib-path],
30 [AS_HELP_STRING([--with-yamlcpp-lib-path], [location of the yamlcpp libraries])],
31 [YAML_LIBS="-L$withval -lyaml-cpp"],
32 [YAML_LIBS='-lyaml-cpp'])
33AC_SUBST([YAML_LIBS])
34
7AC_OUTPUT 35AC_OUTPUT
diff --git a/kgramstats.cpp b/kgramstats.cpp new file mode 100644 index 0000000..142b5aa --- /dev/null +++ b/kgramstats.cpp
@@ -0,0 +1,110 @@
1#include "kgramstats.h"
2#include <vector>
3#include <iostream>
4#include <cstdlib>
5
6kgramstats::kgramstats(string corpus, int maxK)
7{
8 this->maxK = maxK;
9
10 vector<string> tokens;
11 int start = 0;
12 int end = 0;
13
14 while (end != string::npos)
15 {
16 end = corpus.find(" ", start);
17
18 tokens.push_back(corpus.substr(start, (end == string::npos) ? string::npos : end - start));
19
20 start = ((end > (string::npos - 1) ) ? string::npos : end + 1);
21 }
22
23 stats = new map<kgram, map<string, int>* >();
24 for (int k=0; k<=maxK; k++)
25 {
26 for (int i=0; i<(tokens.size() - k); i++)
27 {
28 kgram seq(tokens.begin()+i, tokens.begin()+i+k);
29 string f = tokens[i+k];
30
31 if ((*stats)[seq] == NULL)
32 {
33 (*stats)[seq] = new map<string, int>();
34 }
35
36 (*((*stats)[seq]))[f]++;
37 }
38 }
39}
40
41map<string, int>* kgramstats::lookupExts(kgram tk)
42{
43 return (*stats)[tk];
44}
45
46int kgramstats::getMaxK()
47{
48 return maxK;
49}
50
51void printKgram(kgram k)
52{
53 for (kgram::iterator it = k.begin(); it != k.end(); it++)
54 {
55 cout << *it << " ";
56 }
57 cout << endl;
58}
59
60vector<string> kgramstats::randomSentence(int n)
61{
62 vector<string> result;
63 list<string> cur;
64
65 for (int i=0; i<n; i++)
66 {
67 if ((rand() % 4) != 0)
68 {
69 for (int i=0; i<cur.size(); i++)
70 {
71 if ((rand() % 3) != 0)
72 {
73 cur.pop_front();
74 } else {
75 break;
76 }
77 }
78 }
79
80 map<string, int>* probtable = lookupExts(cur);
81 int max = 0;
82 for (map<string, int>::iterator it = probtable->begin(); it != probtable->end(); ++it)
83 {
84 max += it->second;
85 }
86
87 int r = rand() % (max+1);
88 string next = probtable->begin()->first;
89 for (map<string, int>::iterator it = probtable->begin(); it != probtable->end(); ++it)
90 {
91 if (it->second > r)
92 {
93 break;
94 } else {
95 next = it->first;
96 r -= it->second;
97 }
98 }
99
100 if (cur.size() == maxK)
101 {
102 cur.pop_front();
103 }
104
105 cur.push_back(next);
106 result.push_back(next);
107 }
108
109 return result;
110} \ No newline at end of file
diff --git a/kgramstats.h b/kgramstats.h new file mode 100644 index 0000000..069bb90 --- /dev/null +++ b/kgramstats.h
@@ -0,0 +1,28 @@
1#include <string>
2#include <map>
3#include <list>
4#include <vector>
5
6using namespace::std;
7
8#ifndef KGRAMSTATS_H
9#define KGRAMSTATS_H
10
11typedef list<string> kgram;
12
13class kgramstats
14{
15public:
16 kgramstats(string corpus, int maxK);
17 map<string, int>* lookupExts(kgram tk);
18 int getMaxK();
19 vector<string> randomSentence(int n);
20
21private:
22 int maxK;
23 map<kgram, map<string, int>* >* stats;
24};
25
26void printKgram(kgram k);
27
28#endif \ No newline at end of file
diff --git a/main.cpp b/main.cpp index 8185573..8310bf4 100644 --- a/main.cpp +++ b/main.cpp
@@ -1,6 +1,167 @@
1#include <cstdio> 1#include <cstdio>
2#include <list>
3#include <map>
4#include "kgramstats.h"
5#include <ctime>
6#include <vector>
7#include <cstdlib>
8#include <fstream>
9#include <iostream>
10#include <twitcurl.h>
11#include <unistd.h>
12#include <yaml-cpp/yaml.h>
13
14using namespace::std;
2 15
3int main(int argc, char** args) 16int main(int argc, char** args)
4{ 17{
18 srand(time(NULL));
19
20 YAML::Node config = YAML::LoadFile("config.yml");
21
22 for (;;)
23 {
24 ifstream infile(config["corpus"].as<std::string>().c_str());
25 string corpus;
26 string line;
27 while (getline(infile, line))
28 {
29 corpus += " " + line;
30 }
31
32 kgramstats* stats = new kgramstats(corpus, 5);
33 vector<string> doc = stats->randomSentence(rand() % 25 + 5);
34 string hi;
35 for (vector<string>::iterator it = doc.begin(); it != doc.end(); ++it)
36 {
37 hi += *it + " ";
38 }
39
40 hi = hi.substr(0,140);
41
42 twitCurl twitterObj;
43 std::string tmpStr, tmpStr2;
44 std::string replyMsg;
45 char tmpBuf[1024];
46 std::string username(config["username"].as<std::string>());
47 std::string password(config["password"].as<std::string>());
48
49 /* Set twitter username and password */
50 twitterObj.setTwitterUsername(username);
51 twitterObj.setTwitterPassword(password);
52
53 /* OAuth flow begins */
54 /* Step 0: Set OAuth related params. These are got by registering your app at twitter.com */
55 twitterObj.getOAuth().setConsumerKey( config["consumer_key"].as<std::string>() );
56 twitterObj.getOAuth().setConsumerSecret( config["consumer_secret"].as<std::string>() );
57
58 /* Step 1: Check if we alredy have OAuth access token from a previous run */
59 std::string myOAuthAccessTokenKey("");
60 std::string myOAuthAccessTokenSecret("");
61 std::ifstream oAuthTokenKeyIn;
62 std::ifstream oAuthTokenSecretIn;
63
64 oAuthTokenKeyIn.open( "twitterClient_token_key.txt" );
65 oAuthTokenSecretIn.open( "twitterClient_token_secret.txt" );
66
67 memset( tmpBuf, 0, 1024 );
68 oAuthTokenKeyIn >> tmpBuf;
69 myOAuthAccessTokenKey = tmpBuf;
70
71 memset( tmpBuf, 0, 1024 );
72 oAuthTokenSecretIn >> tmpBuf;
73 myOAuthAccessTokenSecret = tmpBuf;
74
75 oAuthTokenKeyIn.close();
76 oAuthTokenSecretIn.close();
77
78 if( myOAuthAccessTokenKey.size() && myOAuthAccessTokenSecret.size() )
79 {
80 /* If we already have these keys, then no need to go through auth again */
81 printf( "\nUsing:\nKey: %s\nSecret: %s\n\n", myOAuthAccessTokenKey.c_str(), myOAuthAccessTokenSecret.c_str() );
82
83 twitterObj.getOAuth().setOAuthTokenKey( myOAuthAccessTokenKey );
84 twitterObj.getOAuth().setOAuthTokenSecret( myOAuthAccessTokenSecret );
85 }
86 else
87 {
88 /* Step 2: Get request token key and secret */
89 std::string authUrl;
90 twitterObj.oAuthRequestToken( authUrl );
91
92 /* Step 3: Get PIN */
93 memset( tmpBuf, 0, 1024 );
94 printf( "\nDo you want to visit twitter.com for PIN (0 for no; 1 for yes): " );
95 gets( tmpBuf );
96 tmpStr = tmpBuf;
97 if( std::string::npos != tmpStr.find( "1" ) )
98 {
99 /* Ask user to visit twitter.com auth page and get PIN */
100 memset( tmpBuf, 0, 1024 );
101 printf( "\nPlease visit this link in web browser and authorize this application:\n%s", authUrl.c_str() );
102 printf( "\nEnter the PIN provided by twitter: " );
103 gets( tmpBuf );
104 tmpStr = tmpBuf;
105 twitterObj.getOAuth().setOAuthPin( tmpStr );
106 }
107 else
108 {
109 /* Else, pass auth url to twitCurl and get it via twitCurl PIN handling */
110 twitterObj.oAuthHandlePIN( authUrl );
111 }
112
113 /* Step 4: Exchange request token with access token */
114 twitterObj.oAuthAccessToken();
115
116 /* Step 5: Now, save this access token key and secret for future use without PIN */
117 twitterObj.getOAuth().getOAuthTokenKey( myOAuthAccessTokenKey );
118 twitterObj.getOAuth().getOAuthTokenSecret( myOAuthAccessTokenSecret );
119
120 /* Step 6: Save these keys in a file or wherever */
121 std::ofstream oAuthTokenKeyOut;
122 std::ofstream oAuthTokenSecretOut;
123
124 oAuthTokenKeyOut.open( "twitterClient_token_key.txt" );
125 oAuthTokenSecretOut.open( "twitterClient_token_secret.txt" );
126
127 oAuthTokenKeyOut.clear();
128 oAuthTokenSecretOut.clear();
129
130 oAuthTokenKeyOut << myOAuthAccessTokenKey.c_str();
131 oAuthTokenSecretOut << myOAuthAccessTokenSecret.c_str();
132
133 oAuthTokenKeyOut.close();
134 oAuthTokenSecretOut.close();
135 }
136 /* OAuth flow ends */
137
138 /* Account credentials verification */
139 if( twitterObj.accountVerifyCredGet() )
140 {
141 twitterObj.getLastWebResponse( replyMsg );
142 printf( "\ntwitterClient:: twitCurl::accountVerifyCredGet web response:\n%s\n", replyMsg.c_str() );
143 }
144 else
145 {
146 twitterObj.getLastCurlError( replyMsg );
147 printf( "\ntwitterClient:: twitCurl::accountVerifyCredGet error:\n%s\n", replyMsg.c_str() );
148 }
149
150 /* Post a new status message */
151 replyMsg = "";
152 if( twitterObj.statusUpdate( hi ) )
153 {
154 twitterObj.getLastWebResponse( replyMsg );
155 printf( "\ntwitterClient:: twitCurl::statusUpdate web response:\n%s\n", replyMsg.c_str() );
156 }
157 else
158 {
159 twitterObj.getLastCurlError( replyMsg );
160 printf( "\ntwitterClient:: twitCurl::statusUpdate error:\n%s\n", replyMsg.c_str() );
161 }
162
163 sleep(900);
164 }
165
5 return 0; 166 return 0;
6} \ No newline at end of file 167} \ No newline at end of file
generated by cgit-pink 1.4.1 (git 2.36.1) at 2024-11-24 21:08:09 +0000