sunscraper 1.0.0 → 1.1.0.beta1
Sign up to get free protection for your applications and to get access to all the features.
- data/.travis.yml +28 -0
- data/README.md +42 -20
- data/ext/.gitignore +5 -1
- data/ext/embed/sunscraper.cpp +92 -0
- data/ext/{sunscraper.h → embed/sunscraper.h} +9 -7
- data/ext/{sunscraper.pro → embed/sunscraper.pro} +2 -0
- data/ext/embed/sunscraperexternal.cpp +39 -0
- data/ext/{sunscraperlibrary.cpp → embed/sunscraperlibrary.cpp} +4 -9
- data/ext/{sunscraperlibrary.h → embed/sunscraperlibrary.h} +1 -5
- data/ext/embed/sunscraperproxy.cpp +14 -0
- data/ext/{sunscraperproxy.h → embed/sunscraperproxy.h} +3 -3
- data/ext/embed/sunscraperthread.cpp +148 -0
- data/ext/embed/sunscraperthread.h +54 -0
- data/ext/extconf.rb +13 -3
- data/ext/standalone/sunscraper.pro +13 -0
- data/ext/standalone/sunscrapermain.cpp +13 -0
- data/ext/{sunscraperproxy.cpp → standalone/sunscraperproxy.cpp} +2 -2
- data/ext/standalone/sunscraperproxy.h +24 -0
- data/ext/standalone/sunscraperrpc.cpp +183 -0
- data/ext/standalone/sunscraperrpc.h +64 -0
- data/ext/{sunscraperthread.cpp → standalone/sunscraperworker.cpp} +9 -18
- data/ext/{sunscraperthread.h → standalone/sunscraperworker.h} +8 -8
- data/lib/sunscraper/library.rb +33 -29
- data/lib/sunscraper/standalone.rb +168 -0
- data/lib/sunscraper.rb +48 -11
- data/spec/sunscraper_spec.rb +59 -13
- data/sunscraper.gemspec +2 -2
- metadata +58 -75
- data/ext/Makefile +0 -270
- data/ext/sunscraper.cpp +0 -86
- data/ext/sunscraperexternal.cpp +0 -33
data/spec/sunscraper_spec.rb
CHANGED
@@ -21,25 +21,71 @@ HTML
|
|
21
21
|
|
22
22
|
PORT = 45555
|
23
23
|
|
24
|
-
|
25
|
-
|
26
|
-
|
24
|
+
def with_webserver
|
25
|
+
server = WEBrick::HTTPServer.new :Port => PORT, :Logger => WEBrick::Log.new('/dev/null'), :AccessLog => []
|
26
|
+
server.mount_proc '/' do |req, res|
|
27
|
+
res.body = HTML
|
27
28
|
end
|
29
|
+
Thread.new { server.start }
|
28
30
|
|
29
|
-
|
30
|
-
|
31
|
-
|
32
|
-
|
31
|
+
yield PORT
|
32
|
+
ensure
|
33
|
+
server.shutdown if server
|
34
|
+
end
|
35
|
+
|
36
|
+
class String
|
37
|
+
def to_v
|
38
|
+
split(".").map(&:to_i).extend Comparable
|
39
|
+
end
|
40
|
+
end
|
41
|
+
|
42
|
+
unless Sunscraper.os_x?
|
43
|
+
# This part currently crashes on OS X (and will forever).
|
44
|
+
describe "Sunscraper::Library" do
|
45
|
+
before do
|
46
|
+
Sunscraper.worker = :embed
|
33
47
|
end
|
34
|
-
Thread.new { server.start }
|
35
48
|
|
36
|
-
|
49
|
+
it "can scrape an HTML provided as a string" do
|
50
|
+
Sunscraper.scrape_html(HTML).should include('It works!')
|
51
|
+
end
|
37
52
|
|
38
|
-
|
53
|
+
it "can scrape an URL" do
|
54
|
+
with_webserver do |port|
|
55
|
+
Sunscraper.scrape_url("http://127.0.0.1:#{port}/").should include('It works!')
|
56
|
+
end
|
57
|
+
end
|
58
|
+
|
59
|
+
it "should time out if callback is not called" do
|
60
|
+
lambda { Sunscraper.scrape_html("<!-- nothing. at least no callbacks -->", 1000) }.
|
61
|
+
should raise_exception(Sunscraper::ScrapeTimeout)
|
62
|
+
end
|
39
63
|
end
|
64
|
+
end
|
40
65
|
|
41
|
-
|
42
|
-
|
43
|
-
|
66
|
+
if !(RUBY_ENGINE =~ /rbx/ || RUBY_ENGINE =~ /jruby/) ||
|
67
|
+
ENV['EXPERIMENTAL'] == 'true'
|
68
|
+
# This part currently crashes Rubinius (as of Mar 09, 2012),
|
69
|
+
# and crashes jruby < 1.7.0, and uses Unix sockets which don't
|
70
|
+
# work even on jruby master (as of Mar 09, 2012).
|
71
|
+
describe "Sunscraper::Standalone" do
|
72
|
+
before do
|
73
|
+
Sunscraper.worker = :standalone
|
74
|
+
end
|
75
|
+
|
76
|
+
it "can scrape an HTML provided as a string" do
|
77
|
+
Sunscraper.scrape_html(HTML).should include('It works!')
|
78
|
+
end
|
79
|
+
|
80
|
+
it "can scrape an URL" do
|
81
|
+
with_webserver do |port|
|
82
|
+
Sunscraper.scrape_url("http://127.0.0.1:#{port}/").should include('It works!')
|
83
|
+
end
|
84
|
+
end
|
85
|
+
|
86
|
+
it "should time out if callback is not called" do
|
87
|
+
lambda { Sunscraper.scrape_html("<!-- nothing. at least no callbacks -->", 1000) }.
|
88
|
+
should raise_exception(Sunscraper::ScrapeTimeout)
|
89
|
+
end
|
44
90
|
end
|
45
91
|
end
|
data/sunscraper.gemspec
CHANGED
@@ -3,10 +3,10 @@ $:.push File.expand_path("../lib", __FILE__)
|
|
3
3
|
|
4
4
|
Gem::Specification.new do |s|
|
5
5
|
s.name = "sunscraper"
|
6
|
-
s.version = "1.0.
|
6
|
+
s.version = "1.1.0.beta1"
|
7
7
|
s.authors = ["Peter Zotov"]
|
8
8
|
s.email = ["whitequark@whitequark.org"]
|
9
|
-
s.homepage = "http://github.com/
|
9
|
+
s.homepage = "http://github.com/whitequark/sunscraper"
|
10
10
|
s.summary = %q{A WebKit-based, JavaScript-capable HTML scraper.}
|
11
11
|
s.description = s.summary
|
12
12
|
|
metadata
CHANGED
@@ -1,120 +1,103 @@
|
|
1
|
-
--- !ruby/object:Gem::Specification
|
1
|
+
--- !ruby/object:Gem::Specification
|
2
2
|
name: sunscraper
|
3
|
-
version: !ruby/object:Gem::Version
|
4
|
-
|
5
|
-
prerelease:
|
6
|
-
segments:
|
7
|
-
- 1
|
8
|
-
- 0
|
9
|
-
- 0
|
10
|
-
version: 1.0.0
|
3
|
+
version: !ruby/object:Gem::Version
|
4
|
+
version: 1.1.0.beta1
|
5
|
+
prerelease: 6
|
11
6
|
platform: ruby
|
12
|
-
authors:
|
7
|
+
authors:
|
13
8
|
- Peter Zotov
|
14
9
|
autorequire:
|
15
10
|
bindir: bin
|
16
11
|
cert_chain: []
|
17
|
-
|
18
|
-
|
19
|
-
|
20
|
-
- !ruby/object:Gem::Dependency
|
12
|
+
date: 2012-03-09 00:00:00.000000000 Z
|
13
|
+
dependencies:
|
14
|
+
- !ruby/object:Gem::Dependency
|
21
15
|
name: rspec
|
22
|
-
|
23
|
-
requirement: &id001 !ruby/object:Gem::Requirement
|
16
|
+
requirement: &80239320 !ruby/object:Gem::Requirement
|
24
17
|
none: false
|
25
|
-
requirements:
|
26
|
-
- -
|
27
|
-
- !ruby/object:Gem::Version
|
28
|
-
|
29
|
-
segments:
|
30
|
-
- 0
|
31
|
-
version: "0"
|
18
|
+
requirements:
|
19
|
+
- - ! '>='
|
20
|
+
- !ruby/object:Gem::Version
|
21
|
+
version: '0'
|
32
22
|
type: :development
|
33
|
-
version_requirements: *id001
|
34
|
-
- !ruby/object:Gem::Dependency
|
35
|
-
name: ffi
|
36
23
|
prerelease: false
|
37
|
-
|
24
|
+
version_requirements: *80239320
|
25
|
+
- !ruby/object:Gem::Dependency
|
26
|
+
name: ffi
|
27
|
+
requirement: &80239070 !ruby/object:Gem::Requirement
|
38
28
|
none: false
|
39
|
-
requirements:
|
40
|
-
- -
|
41
|
-
- !ruby/object:Gem::Version
|
42
|
-
hash: 716237343
|
43
|
-
segments:
|
44
|
-
- 1
|
45
|
-
- 0
|
46
|
-
- 11
|
29
|
+
requirements:
|
30
|
+
- - ! '>='
|
31
|
+
- !ruby/object:Gem::Version
|
47
32
|
version: 1.0.11
|
48
33
|
type: :runtime
|
49
|
-
|
34
|
+
prerelease: false
|
35
|
+
version_requirements: *80239070
|
50
36
|
description: A WebKit-based, JavaScript-capable HTML scraper.
|
51
|
-
email:
|
37
|
+
email:
|
52
38
|
- whitequark@whitequark.org
|
53
39
|
executables: []
|
54
|
-
|
55
|
-
extensions:
|
40
|
+
extensions:
|
56
41
|
- ext/extconf.rb
|
57
42
|
extra_rdoc_files: []
|
58
|
-
|
59
|
-
files:
|
43
|
+
files:
|
60
44
|
- .gitignore
|
61
45
|
- .rspec
|
46
|
+
- .travis.yml
|
62
47
|
- .yardopts
|
63
48
|
- Gemfile
|
64
49
|
- LICENSE
|
65
50
|
- README.md
|
66
51
|
- Rakefile
|
67
52
|
- ext/.gitignore
|
68
|
-
- ext/
|
53
|
+
- ext/embed/sunscraper.cpp
|
54
|
+
- ext/embed/sunscraper.h
|
55
|
+
- ext/embed/sunscraper.pro
|
56
|
+
- ext/embed/sunscraperexternal.cpp
|
57
|
+
- ext/embed/sunscraperlibrary.cpp
|
58
|
+
- ext/embed/sunscraperlibrary.h
|
59
|
+
- ext/embed/sunscraperproxy.cpp
|
60
|
+
- ext/embed/sunscraperproxy.h
|
61
|
+
- ext/embed/sunscraperthread.cpp
|
62
|
+
- ext/embed/sunscraperthread.h
|
69
63
|
- ext/extconf.rb
|
70
|
-
- ext/sunscraper.
|
71
|
-
- ext/
|
72
|
-
- ext/
|
73
|
-
- ext/
|
74
|
-
- ext/
|
75
|
-
- ext/
|
76
|
-
- ext/
|
77
|
-
- ext/
|
78
|
-
- ext/sunscraperthread.cpp
|
79
|
-
- ext/sunscraperthread.h
|
64
|
+
- ext/standalone/sunscraper.pro
|
65
|
+
- ext/standalone/sunscrapermain.cpp
|
66
|
+
- ext/standalone/sunscraperproxy.cpp
|
67
|
+
- ext/standalone/sunscraperproxy.h
|
68
|
+
- ext/standalone/sunscraperrpc.cpp
|
69
|
+
- ext/standalone/sunscraperrpc.h
|
70
|
+
- ext/standalone/sunscraperworker.cpp
|
71
|
+
- ext/standalone/sunscraperworker.h
|
80
72
|
- lib/sunscraper.rb
|
81
73
|
- lib/sunscraper/library.rb
|
74
|
+
- lib/sunscraper/standalone.rb
|
82
75
|
- spec/spec_helper.rb
|
83
76
|
- spec/sunscraper_spec.rb
|
84
77
|
- sunscraper.gemspec
|
85
|
-
homepage: http://github.com/
|
78
|
+
homepage: http://github.com/whitequark/sunscraper
|
86
79
|
licenses: []
|
87
|
-
|
88
80
|
post_install_message:
|
89
81
|
rdoc_options: []
|
90
|
-
|
91
|
-
require_paths:
|
82
|
+
require_paths:
|
92
83
|
- lib
|
93
|
-
required_ruby_version: !ruby/object:Gem::Requirement
|
84
|
+
required_ruby_version: !ruby/object:Gem::Requirement
|
94
85
|
none: false
|
95
|
-
requirements:
|
96
|
-
- -
|
97
|
-
- !ruby/object:Gem::Version
|
98
|
-
|
99
|
-
|
100
|
-
- 0
|
101
|
-
version: "0"
|
102
|
-
required_rubygems_version: !ruby/object:Gem::Requirement
|
86
|
+
requirements:
|
87
|
+
- - ! '>='
|
88
|
+
- !ruby/object:Gem::Version
|
89
|
+
version: '0'
|
90
|
+
required_rubygems_version: !ruby/object:Gem::Requirement
|
103
91
|
none: false
|
104
|
-
requirements:
|
105
|
-
- -
|
106
|
-
- !ruby/object:Gem::Version
|
107
|
-
|
108
|
-
segments:
|
109
|
-
- 0
|
110
|
-
version: "0"
|
92
|
+
requirements:
|
93
|
+
- - ! '>'
|
94
|
+
- !ruby/object:Gem::Version
|
95
|
+
version: 1.3.1
|
111
96
|
requirements: []
|
112
|
-
|
113
97
|
rubyforge_project: sunscraper
|
114
|
-
rubygems_version: 1.8.
|
98
|
+
rubygems_version: 1.8.17
|
115
99
|
signing_key:
|
116
100
|
specification_version: 3
|
117
101
|
summary: A WebKit-based, JavaScript-capable HTML scraper.
|
118
102
|
test_files: []
|
119
|
-
|
120
103
|
has_rdoc:
|
data/ext/Makefile
DELETED
@@ -1,270 +0,0 @@
|
|
1
|
-
#############################################################################
|
2
|
-
# Makefile for building: libsunscraper.so.1.0.0
|
3
|
-
# Generated by qmake (2.01a) (Qt 4.7.4) on: Sat Feb 18 05:54:02 2012
|
4
|
-
# Project: sunscraper.pro
|
5
|
-
# Template: lib
|
6
|
-
# Command: /usr/bin/qmake-qt4 -spec /usr/share/qt4/mkspecs/linux-g++ CONFIG+=debug -o Makefile sunscraper.pro
|
7
|
-
#############################################################################
|
8
|
-
|
9
|
-
####### Compiler, tools and options
|
10
|
-
|
11
|
-
CC = gcc
|
12
|
-
CXX = g++
|
13
|
-
DEFINES = -DQT_WEBKIT -DQT_WEBKIT_LIB -DQT_GUI_LIB -DQT_CORE_LIB -DQT_SHARED
|
14
|
-
CFLAGS = -pipe -g -Wall -W -D_REENTRANT -fPIC $(DEFINES)
|
15
|
-
CXXFLAGS = -pipe -g -Wall -W -D_REENTRANT -fPIC $(DEFINES)
|
16
|
-
INCPATH = -I/usr/share/qt4/mkspecs/linux-g++ -I. -I/usr/include/qt4/QtCore -I/usr/include/qt4/QtGui -I/usr/include/qt4/QtWebKit -I/usr/include/qt4 -I.
|
17
|
-
LINK = g++
|
18
|
-
LFLAGS = -shared -Wl,-soname,libsunscraper.so.1
|
19
|
-
LIBS = $(SUBLIBS) -L/usr/lib -lQtWebKit -lQtGui -lQtCore -lpthread
|
20
|
-
AR = ar cqs
|
21
|
-
RANLIB =
|
22
|
-
QMAKE = /usr/bin/qmake-qt4
|
23
|
-
TAR = tar -cf
|
24
|
-
COMPRESS = gzip -9f
|
25
|
-
COPY = cp -f
|
26
|
-
SED = sed
|
27
|
-
COPY_FILE = $(COPY)
|
28
|
-
COPY_DIR = $(COPY) -r
|
29
|
-
STRIP = strip
|
30
|
-
INSTALL_FILE = install -m 644 -p
|
31
|
-
INSTALL_DIR = $(COPY_DIR)
|
32
|
-
INSTALL_PROGRAM = install -m 755 -p
|
33
|
-
DEL_FILE = rm -f
|
34
|
-
SYMLINK = ln -f -s
|
35
|
-
DEL_DIR = rmdir
|
36
|
-
MOVE = mv -f
|
37
|
-
CHK_DIR_EXISTS= test -d
|
38
|
-
MKDIR = mkdir -p
|
39
|
-
|
40
|
-
####### Output directory
|
41
|
-
|
42
|
-
OBJECTS_DIR = ./
|
43
|
-
|
44
|
-
####### Files
|
45
|
-
|
46
|
-
SOURCES = sunscraperlibrary.cpp \
|
47
|
-
sunscraperthread.cpp \
|
48
|
-
sunscraperexternal.cpp \
|
49
|
-
sunscraper.cpp \
|
50
|
-
sunscraperproxy.cpp moc_sunscraperthread.cpp \
|
51
|
-
moc_sunscraper.cpp \
|
52
|
-
moc_sunscraperproxy.cpp
|
53
|
-
OBJECTS = sunscraperlibrary.o \
|
54
|
-
sunscraperthread.o \
|
55
|
-
sunscraperexternal.o \
|
56
|
-
sunscraper.o \
|
57
|
-
sunscraperproxy.o \
|
58
|
-
moc_sunscraperthread.o \
|
59
|
-
moc_sunscraper.o \
|
60
|
-
moc_sunscraperproxy.o
|
61
|
-
DIST = /usr/share/qt4/mkspecs/common/g++.conf \
|
62
|
-
/usr/share/qt4/mkspecs/common/unix.conf \
|
63
|
-
/usr/share/qt4/mkspecs/common/linux.conf \
|
64
|
-
/usr/share/qt4/mkspecs/qconfig.pri \
|
65
|
-
/usr/share/qt4/mkspecs/modules/qt_webkit_version.pri \
|
66
|
-
/usr/share/qt4/mkspecs/features/qt_functions.prf \
|
67
|
-
/usr/share/qt4/mkspecs/features/qt_config.prf \
|
68
|
-
/usr/share/qt4/mkspecs/features/exclusive_builds.prf \
|
69
|
-
/usr/share/qt4/mkspecs/features/default_pre.prf \
|
70
|
-
/usr/share/qt4/mkspecs/features/debug.prf \
|
71
|
-
/usr/share/qt4/mkspecs/features/default_post.prf \
|
72
|
-
/usr/share/qt4/mkspecs/features/warn_on.prf \
|
73
|
-
/usr/share/qt4/mkspecs/features/qt.prf \
|
74
|
-
/usr/share/qt4/mkspecs/features/unix/thread.prf \
|
75
|
-
/usr/share/qt4/mkspecs/features/moc.prf \
|
76
|
-
/usr/share/qt4/mkspecs/features/resources.prf \
|
77
|
-
/usr/share/qt4/mkspecs/features/uic.prf \
|
78
|
-
/usr/share/qt4/mkspecs/features/yacc.prf \
|
79
|
-
/usr/share/qt4/mkspecs/features/lex.prf \
|
80
|
-
/usr/share/qt4/mkspecs/features/include_source_dir.prf \
|
81
|
-
sunscraper.pro
|
82
|
-
QMAKE_TARGET = sunscraper
|
83
|
-
DESTDIR =
|
84
|
-
TARGET = libsunscraper.so.1.0.0
|
85
|
-
TARGETA = libsunscraper.a
|
86
|
-
TARGETD = libsunscraper.so.1.0.0
|
87
|
-
TARGET0 = libsunscraper.so
|
88
|
-
TARGET1 = libsunscraper.so.1
|
89
|
-
TARGET2 = libsunscraper.so.1.0
|
90
|
-
|
91
|
-
first: all
|
92
|
-
####### Implicit rules
|
93
|
-
|
94
|
-
.SUFFIXES: .o .c .cpp .cc .cxx .C
|
95
|
-
|
96
|
-
.cpp.o:
|
97
|
-
$(CXX) -c $(CXXFLAGS) $(INCPATH) -o "$@" "$<"
|
98
|
-
|
99
|
-
.cc.o:
|
100
|
-
$(CXX) -c $(CXXFLAGS) $(INCPATH) -o "$@" "$<"
|
101
|
-
|
102
|
-
.cxx.o:
|
103
|
-
$(CXX) -c $(CXXFLAGS) $(INCPATH) -o "$@" "$<"
|
104
|
-
|
105
|
-
.C.o:
|
106
|
-
$(CXX) -c $(CXXFLAGS) $(INCPATH) -o "$@" "$<"
|
107
|
-
|
108
|
-
.c.o:
|
109
|
-
$(CC) -c $(CFLAGS) $(INCPATH) -o "$@" "$<"
|
110
|
-
|
111
|
-
####### Build rules
|
112
|
-
|
113
|
-
all: Makefile $(TARGET)
|
114
|
-
|
115
|
-
$(TARGET): $(OBJECTS) $(SUBLIBS) $(OBJCOMP)
|
116
|
-
-$(DEL_FILE) $(TARGET) $(TARGET0) $(TARGET1) $(TARGET2)
|
117
|
-
$(LINK) $(LFLAGS) -o $(TARGET) $(OBJECTS) $(LIBS) $(OBJCOMP)
|
118
|
-
-ln -s $(TARGET) $(TARGET0)
|
119
|
-
-ln -s $(TARGET) $(TARGET1)
|
120
|
-
-ln -s $(TARGET) $(TARGET2)
|
121
|
-
|
122
|
-
|
123
|
-
|
124
|
-
staticlib: $(TARGETA)
|
125
|
-
|
126
|
-
$(TARGETA): $(OBJECTS) $(OBJCOMP)
|
127
|
-
-$(DEL_FILE) $(TARGETA)
|
128
|
-
$(AR) $(TARGETA) $(OBJECTS)
|
129
|
-
|
130
|
-
Makefile: sunscraper.pro /usr/share/qt4/mkspecs/linux-g++/qmake.conf /usr/share/qt4/mkspecs/common/g++.conf \
|
131
|
-
/usr/share/qt4/mkspecs/common/unix.conf \
|
132
|
-
/usr/share/qt4/mkspecs/common/linux.conf \
|
133
|
-
/usr/share/qt4/mkspecs/qconfig.pri \
|
134
|
-
/usr/share/qt4/mkspecs/modules/qt_webkit_version.pri \
|
135
|
-
/usr/share/qt4/mkspecs/features/qt_functions.prf \
|
136
|
-
/usr/share/qt4/mkspecs/features/qt_config.prf \
|
137
|
-
/usr/share/qt4/mkspecs/features/exclusive_builds.prf \
|
138
|
-
/usr/share/qt4/mkspecs/features/default_pre.prf \
|
139
|
-
/usr/share/qt4/mkspecs/features/debug.prf \
|
140
|
-
/usr/share/qt4/mkspecs/features/default_post.prf \
|
141
|
-
/usr/share/qt4/mkspecs/features/warn_on.prf \
|
142
|
-
/usr/share/qt4/mkspecs/features/qt.prf \
|
143
|
-
/usr/share/qt4/mkspecs/features/unix/thread.prf \
|
144
|
-
/usr/share/qt4/mkspecs/features/moc.prf \
|
145
|
-
/usr/share/qt4/mkspecs/features/resources.prf \
|
146
|
-
/usr/share/qt4/mkspecs/features/uic.prf \
|
147
|
-
/usr/share/qt4/mkspecs/features/yacc.prf \
|
148
|
-
/usr/share/qt4/mkspecs/features/lex.prf \
|
149
|
-
/usr/share/qt4/mkspecs/features/include_source_dir.prf \
|
150
|
-
/usr/lib/libQtWebKit.prl \
|
151
|
-
/usr/lib/libQtGui.prl \
|
152
|
-
/usr/lib/libQtCore.prl
|
153
|
-
$(QMAKE) -spec /usr/share/qt4/mkspecs/linux-g++ CONFIG+=debug -o Makefile sunscraper.pro
|
154
|
-
/usr/share/qt4/mkspecs/common/g++.conf:
|
155
|
-
/usr/share/qt4/mkspecs/common/unix.conf:
|
156
|
-
/usr/share/qt4/mkspecs/common/linux.conf:
|
157
|
-
/usr/share/qt4/mkspecs/qconfig.pri:
|
158
|
-
/usr/share/qt4/mkspecs/modules/qt_webkit_version.pri:
|
159
|
-
/usr/share/qt4/mkspecs/features/qt_functions.prf:
|
160
|
-
/usr/share/qt4/mkspecs/features/qt_config.prf:
|
161
|
-
/usr/share/qt4/mkspecs/features/exclusive_builds.prf:
|
162
|
-
/usr/share/qt4/mkspecs/features/default_pre.prf:
|
163
|
-
/usr/share/qt4/mkspecs/features/debug.prf:
|
164
|
-
/usr/share/qt4/mkspecs/features/default_post.prf:
|
165
|
-
/usr/share/qt4/mkspecs/features/warn_on.prf:
|
166
|
-
/usr/share/qt4/mkspecs/features/qt.prf:
|
167
|
-
/usr/share/qt4/mkspecs/features/unix/thread.prf:
|
168
|
-
/usr/share/qt4/mkspecs/features/moc.prf:
|
169
|
-
/usr/share/qt4/mkspecs/features/resources.prf:
|
170
|
-
/usr/share/qt4/mkspecs/features/uic.prf:
|
171
|
-
/usr/share/qt4/mkspecs/features/yacc.prf:
|
172
|
-
/usr/share/qt4/mkspecs/features/lex.prf:
|
173
|
-
/usr/share/qt4/mkspecs/features/include_source_dir.prf:
|
174
|
-
/usr/lib/libQtWebKit.prl:
|
175
|
-
/usr/lib/libQtGui.prl:
|
176
|
-
/usr/lib/libQtCore.prl:
|
177
|
-
qmake: FORCE
|
178
|
-
@$(QMAKE) -spec /usr/share/qt4/mkspecs/linux-g++ CONFIG+=debug -o Makefile sunscraper.pro
|
179
|
-
|
180
|
-
dist:
|
181
|
-
@$(CHK_DIR_EXISTS) .tmp/sunscraper1.0.0 || $(MKDIR) .tmp/sunscraper1.0.0
|
182
|
-
$(COPY_FILE) --parents $(SOURCES) $(DIST) .tmp/sunscraper1.0.0/ && $(COPY_FILE) --parents sunscraperlibrary.h sunscraperthread.h sunscraper.h sunscraperproxy.h .tmp/sunscraper1.0.0/ && $(COPY_FILE) --parents sunscraperlibrary.cpp sunscraperthread.cpp sunscraperexternal.cpp sunscraper.cpp sunscraperproxy.cpp .tmp/sunscraper1.0.0/ && (cd `dirname .tmp/sunscraper1.0.0` && $(TAR) sunscraper1.0.0.tar sunscraper1.0.0 && $(COMPRESS) sunscraper1.0.0.tar) && $(MOVE) `dirname .tmp/sunscraper1.0.0`/sunscraper1.0.0.tar.gz . && $(DEL_FILE) -r .tmp/sunscraper1.0.0
|
183
|
-
|
184
|
-
|
185
|
-
clean:compiler_clean
|
186
|
-
-$(DEL_FILE) $(OBJECTS)
|
187
|
-
-$(DEL_FILE) *~ core *.core
|
188
|
-
|
189
|
-
|
190
|
-
####### Sub-libraries
|
191
|
-
|
192
|
-
distclean: clean
|
193
|
-
-$(DEL_FILE) $(TARGET)
|
194
|
-
-$(DEL_FILE) $(TARGET0) $(TARGET1) $(TARGET2) $(TARGETA)
|
195
|
-
-$(DEL_FILE) Makefile
|
196
|
-
|
197
|
-
|
198
|
-
check: first
|
199
|
-
|
200
|
-
mocclean: compiler_moc_header_clean compiler_moc_source_clean
|
201
|
-
|
202
|
-
mocables: compiler_moc_header_make_all compiler_moc_source_make_all
|
203
|
-
|
204
|
-
compiler_moc_header_make_all: moc_sunscraperthread.cpp moc_sunscraper.cpp moc_sunscraperproxy.cpp
|
205
|
-
compiler_moc_header_clean:
|
206
|
-
-$(DEL_FILE) moc_sunscraperthread.cpp moc_sunscraper.cpp moc_sunscraperproxy.cpp
|
207
|
-
moc_sunscraperthread.cpp: sunscraperthread.h
|
208
|
-
/usr/bin/moc-qt4 $(DEFINES) $(INCPATH) sunscraperthread.h -o moc_sunscraperthread.cpp
|
209
|
-
|
210
|
-
moc_sunscraper.cpp: sunscraper.h
|
211
|
-
/usr/bin/moc-qt4 $(DEFINES) $(INCPATH) sunscraper.h -o moc_sunscraper.cpp
|
212
|
-
|
213
|
-
moc_sunscraperproxy.cpp: sunscraperproxy.h
|
214
|
-
/usr/bin/moc-qt4 $(DEFINES) $(INCPATH) sunscraperproxy.h -o moc_sunscraperproxy.cpp
|
215
|
-
|
216
|
-
compiler_rcc_make_all:
|
217
|
-
compiler_rcc_clean:
|
218
|
-
compiler_image_collection_make_all: qmake_image_collection.cpp
|
219
|
-
compiler_image_collection_clean:
|
220
|
-
-$(DEL_FILE) qmake_image_collection.cpp
|
221
|
-
compiler_moc_source_make_all:
|
222
|
-
compiler_moc_source_clean:
|
223
|
-
compiler_uic_make_all:
|
224
|
-
compiler_uic_clean:
|
225
|
-
compiler_yacc_decl_make_all:
|
226
|
-
compiler_yacc_decl_clean:
|
227
|
-
compiler_yacc_impl_make_all:
|
228
|
-
compiler_yacc_impl_clean:
|
229
|
-
compiler_lex_make_all:
|
230
|
-
compiler_lex_clean:
|
231
|
-
compiler_clean: compiler_moc_header_clean
|
232
|
-
|
233
|
-
####### Compile
|
234
|
-
|
235
|
-
sunscraperlibrary.o: sunscraperlibrary.cpp sunscraperlibrary.h \
|
236
|
-
sunscraperthread.h
|
237
|
-
$(CXX) -c $(CXXFLAGS) $(INCPATH) -o sunscraperlibrary.o sunscraperlibrary.cpp
|
238
|
-
|
239
|
-
sunscraperthread.o: sunscraperthread.cpp sunscraperthread.h \
|
240
|
-
sunscraperproxy.h
|
241
|
-
$(CXX) -c $(CXXFLAGS) $(INCPATH) -o sunscraperthread.o sunscraperthread.cpp
|
242
|
-
|
243
|
-
sunscraperexternal.o: sunscraperexternal.cpp sunscraper.h
|
244
|
-
$(CXX) -c $(CXXFLAGS) $(INCPATH) -o sunscraperexternal.o sunscraperexternal.cpp
|
245
|
-
|
246
|
-
sunscraper.o: sunscraper.cpp sunscraper.h \
|
247
|
-
sunscraperlibrary.h \
|
248
|
-
sunscraperthread.h
|
249
|
-
$(CXX) -c $(CXXFLAGS) $(INCPATH) -o sunscraper.o sunscraper.cpp
|
250
|
-
|
251
|
-
sunscraperproxy.o: sunscraperproxy.cpp sunscraperproxy.h
|
252
|
-
$(CXX) -c $(CXXFLAGS) $(INCPATH) -o sunscraperproxy.o sunscraperproxy.cpp
|
253
|
-
|
254
|
-
moc_sunscraperthread.o: moc_sunscraperthread.cpp
|
255
|
-
$(CXX) -c $(CXXFLAGS) $(INCPATH) -o moc_sunscraperthread.o moc_sunscraperthread.cpp
|
256
|
-
|
257
|
-
moc_sunscraper.o: moc_sunscraper.cpp
|
258
|
-
$(CXX) -c $(CXXFLAGS) $(INCPATH) -o moc_sunscraper.o moc_sunscraper.cpp
|
259
|
-
|
260
|
-
moc_sunscraperproxy.o: moc_sunscraperproxy.cpp
|
261
|
-
$(CXX) -c $(CXXFLAGS) $(INCPATH) -o moc_sunscraperproxy.o moc_sunscraperproxy.cpp
|
262
|
-
|
263
|
-
####### Install
|
264
|
-
|
265
|
-
install: FORCE
|
266
|
-
|
267
|
-
uninstall: FORCE
|
268
|
-
|
269
|
-
FORCE:
|
270
|
-
|
data/ext/sunscraper.cpp
DELETED
@@ -1,86 +0,0 @@
|
|
1
|
-
#include <QApplication>
|
2
|
-
#include <QThread>
|
3
|
-
#include <QTimer>
|
4
|
-
#include <QWebPage>
|
5
|
-
#include <QWebFrame>
|
6
|
-
#include <QMutexLocker>
|
7
|
-
#include <QtDebug>
|
8
|
-
#include "sunscraper.h"
|
9
|
-
#include "sunscraperlibrary.h"
|
10
|
-
#include "sunscraperthread.h"
|
11
|
-
|
12
|
-
unsigned Sunscraper::_nextQueryId = 1;
|
13
|
-
QMutex Sunscraper::_staticMutex;
|
14
|
-
|
15
|
-
Sunscraper::Sunscraper()
|
16
|
-
{
|
17
|
-
QMutexLocker locker(&_staticMutex);
|
18
|
-
|
19
|
-
_queryId = _nextQueryId++;
|
20
|
-
|
21
|
-
SunscraperThread *worker = SunscraperLibrary::instance()->thread();
|
22
|
-
|
23
|
-
connect(this, SIGNAL(requestLoadHtml(uint,QString)),
|
24
|
-
worker, SLOT(loadHtml(uint,QString)), Qt::QueuedConnection);
|
25
|
-
connect(this, SIGNAL(requestLoadUrl(uint,QString)),
|
26
|
-
worker, SLOT(loadUrl(uint,QString)), Qt::QueuedConnection);
|
27
|
-
connect(this, SIGNAL(requestFinalize(uint)),
|
28
|
-
worker, SLOT(finalize(uint)), Qt::QueuedConnection);
|
29
|
-
|
30
|
-
connect(worker, SIGNAL(finished(uint,QString)),
|
31
|
-
this, SLOT(finished(uint,QString)), Qt::QueuedConnection);
|
32
|
-
}
|
33
|
-
|
34
|
-
void Sunscraper::loadHtml(QString html)
|
35
|
-
{
|
36
|
-
emit requestLoadHtml(_queryId, html);
|
37
|
-
}
|
38
|
-
|
39
|
-
void Sunscraper::loadUrl(QString url)
|
40
|
-
{
|
41
|
-
emit requestLoadUrl(_queryId, url);
|
42
|
-
}
|
43
|
-
|
44
|
-
void Sunscraper::wait(unsigned timeout)
|
45
|
-
{
|
46
|
-
QTimer _timeoutTimer;
|
47
|
-
connect(&_timeoutTimer, SIGNAL(timeout()), this, SLOT(timeout()));
|
48
|
-
|
49
|
-
_timeoutTimer.setInterval(timeout);
|
50
|
-
_timeoutTimer.start();
|
51
|
-
|
52
|
-
_eventLoop.exec();
|
53
|
-
|
54
|
-
_timeoutTimer.stop();
|
55
|
-
}
|
56
|
-
|
57
|
-
void Sunscraper::finished(unsigned eventQueryId, QString html)
|
58
|
-
{
|
59
|
-
if(eventQueryId != _queryId)
|
60
|
-
return;
|
61
|
-
|
62
|
-
_eventLoop.quit();
|
63
|
-
|
64
|
-
_html = html.toUtf8();
|
65
|
-
|
66
|
-
emit requestFinalize(_queryId);
|
67
|
-
}
|
68
|
-
|
69
|
-
void Sunscraper::timeout()
|
70
|
-
{
|
71
|
-
_eventLoop.quit();
|
72
|
-
|
73
|
-
_html = "!SUNSCRAPER_TIMEOUT";
|
74
|
-
|
75
|
-
emit requestFinalize(_queryId);
|
76
|
-
}
|
77
|
-
|
78
|
-
QByteArray Sunscraper::fetch()
|
79
|
-
{
|
80
|
-
return _html;
|
81
|
-
}
|
82
|
-
|
83
|
-
const char *Sunscraper::fetchAsCString()
|
84
|
-
{
|
85
|
-
return _html.constData();
|
86
|
-
}
|
data/ext/sunscraperexternal.cpp
DELETED
@@ -1,33 +0,0 @@
|
|
1
|
-
#include "sunscraper.h"
|
2
|
-
|
3
|
-
extern "C" {
|
4
|
-
Sunscraper *sunscraper_create()
|
5
|
-
{
|
6
|
-
return new Sunscraper();
|
7
|
-
}
|
8
|
-
|
9
|
-
void sunscraper_load_html(Sunscraper *sunscraper, const char *html)
|
10
|
-
{
|
11
|
-
sunscraper->loadHtml(html);
|
12
|
-
}
|
13
|
-
|
14
|
-
void sunscraper_load_url(Sunscraper *sunscraper, const char *url)
|
15
|
-
{
|
16
|
-
sunscraper->loadUrl(url);
|
17
|
-
}
|
18
|
-
|
19
|
-
void sunscraper_wait(Sunscraper *sunscraper, unsigned timeout)
|
20
|
-
{
|
21
|
-
sunscraper->wait(timeout);
|
22
|
-
}
|
23
|
-
|
24
|
-
const char *sunscraper_fetch(Sunscraper *sunscraper)
|
25
|
-
{
|
26
|
-
return sunscraper->fetchAsCString();
|
27
|
-
}
|
28
|
-
|
29
|
-
void sunscraper_discard(Sunscraper *sunscraper)
|
30
|
-
{
|
31
|
-
delete sunscraper;
|
32
|
-
}
|
33
|
-
}
|