karchnu
/
mechanize.cr
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Merge pull request #15 from Kanezoh/fix-comment 

Fix comment
master
Kanezoh 2021-11-18 14:22:16 +09:00 committed by GitHub
parent 0ceaffc7b0 ed57c3a696
commit a247432bc4
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
19 changed files with 222 additions and 216 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

346
src/mechanize/form.cr
View File

@ -8,197 +8,199 @@ require "./form/button"
require "./form/select_list"
require "./utils/element_matcher"
# This class represents the form tag of html.
class Mechanize::Form
include Mechanize::ElementMatcher
class Mechanize
# This class represents the form tag of html.
class Form
include ElementMatcher
getter node : Node | Lexbor::Node
# returns hoge array of `Mechanize::FormContent::Field` in the form.
getter fields : Array(FormContent::Field)
# returns an array of input tags whose type is checkbox in the form.
getter checkboxes : Array(FormContent::CheckBox)
# returns an array of input tags whose type is radio in the form.
getter radiobuttons : Array(FormContent::RadioButton)
# returns an array of input tags whose type is select in the form.
getter selectboxes : Array(FormContent::MultiSelectList)
# returns an array of button tags and input tag whose type is button,submit,reset,image.
getter buttons : Array(FormContent::Button)
# returns form's 'enctype' attribute.
getter enctype : String
# returns form's 'method' attribute.
getter method : String
# returns form's 'name' attribute.
getter name : String
# return form's 'action' attribute.
property action : String
# returns the page which includes the form.
getter page : Mechanize::Page?
getter node : Node | Lexbor::Node
# returns hoge array of `Mechanize::FormContent::Field` in the form.
getter fields : Array(FormContent::Field)
# returns an array of input tags whose type is checkbox in the form.
getter checkboxes : Array(FormContent::CheckBox)
# returns an array of input tags whose type is radio in the form.
getter radiobuttons : Array(FormContent::RadioButton)
# returns an array of input tags whose type is select in the form.
getter selectboxes : Array(FormContent::MultiSelectList)
# returns an array of button tags and input tag whose type is button,submit,reset,image.
getter buttons : Array(FormContent::Button)
# returns form's 'enctype' attribute.
getter enctype : String
# returns form's 'method' attribute.
getter method : String
# returns form's 'name' attribute.
getter name : String
# return form's 'action' attribute.
property action : String
# returns the page which includes the form.
getter page : Page?
def initialize(node : Node | Lexbor::Node, page : Mechanize::Page? = nil)
@enctype = node.fetch("enctype", "application/x-www-form-urlencoded")
@node = node
@fields = Array(Mechanize::FormContent::Field).new
@checkboxes = Array(Mechanize::FormContent::CheckBox).new
@radiobuttons = Array(Mechanize::FormContent::RadioButton).new
@selectboxes = Array(Mechanize::FormContent::MultiSelectList).new
@buttons = Array(Mechanize::FormContent::Button).new
@action = node.fetch("action", "")
@method = node.fetch("method", "GET").upcase
@name = node.fetch("name", "")
@clicked_buttons = Array(Mechanize::FormContent::Button).new
@page = page
# @mech = mech
def initialize(node : Node | Lexbor::Node, page : Page? = nil)
@enctype = node.fetch("enctype", "application/x-www-form-urlencoded")
@node = node
@fields = Array(FormContent::Field).new
@checkboxes = Array(FormContent::CheckBox).new
@radiobuttons = Array(FormContent::RadioButton).new
@selectboxes = Array(FormContent::MultiSelectList).new
@buttons = Array(FormContent::Button).new
@action = node.fetch("action", "")
@method = node.fetch("method", "GET").upcase
@name = node.fetch("name", "")
@clicked_buttons = Array(FormContent::Button).new
@page = page
# @mech = mech
# @encoding = node['accept-charset'] || (page && page.encoding) || nil
# @ignore_encoding_error = false
parse
end
# @encoding = node['accept-charset'] || (page && page.encoding) || nil
# @ignore_encoding_error = false
parse
end
def request_data
query_params = build_query
build_query_string(query_params)
end
def request_data
query_params = build_query
build_query_string(query_params)
end
# generate fields_with and field_with methods.
# These methods are used for finding nodes that matches conditions.
# ex.) field_with("email") finds <input name="email">
# generate fields_with and field_with methods.
# These methods are used for finding nodes that matches conditions.
# ex.) field_with("email") finds <input name="email">
elements_with "field"
elements_with "radiobutton"
elements_with "checkbox", "checkboxes"
elements_with "button"
elements_with "field"
elements_with "radiobutton"
elements_with "checkbox", "checkboxes"
elements_with "button"
# Returns all fields of <input type="textarea">
def textareas
fields.select { |f| f.class == FormContent::Textarea }.map &.as(FormContent::Textarea)
end
# Returns all fields of &lt;input type="textarea"&gt;
def textareas
fields.select { |f| f.class == FormContent::Textarea }.map &.as(FormContent::Textarea)
end
private def parse
@node.css("input").not_nil!.each do |html_node|
html_node = html_node.as(Lexbor::Node)
type = (html_node["type"]? || "text").downcase
case type
when "checkbox"
checkboxes << Mechanize::FormContent::CheckBox.new(html_node, self)
when "radio"
radiobuttons << Mechanize::FormContent::RadioButton.new(html_node, self)
when "button"
buttons << Mechanize::FormContent::Button.new(html_node, @node)
when "submit"
buttons << Mechanize::FormContent::SubmitButton.new(html_node, @node)
when "reset"
buttons << Mechanize::FormContent::ResetButton.new(html_node, @node)
when "image"
buttons << Mechanize::FormContent::ImageButton.new(html_node, @node)
when "text"
fields << Mechanize::FormContent::Text.new(html_node)
when "hidden"
fields << Mechanize::FormContent::Hidden.new(html_node)
when "textarea"
fields << Mechanize::FormContent::Textarea.new(html_node)
else
fields << Mechanize::FormContent::Field.new(html_node)
private def parse
@node.css("input").not_nil!.each do |html_node|
html_node = html_node.as(Lexbor::Node)
type = (html_node["type"]? || "text").downcase
case type
when "checkbox"
checkboxes << FormContent::CheckBox.new(html_node, self)
when "radio"
radiobuttons << FormContent::RadioButton.new(html_node, self)
when "button"
buttons << FormContent::Button.new(html_node, @node)
when "submit"
buttons << FormContent::SubmitButton.new(html_node, @node)
when "reset"
buttons << FormContent::ResetButton.new(html_node, @node)
when "image"
buttons << FormContent::ImageButton.new(html_node, @node)
when "text"
fields << FormContent::Text.new(html_node)
when "hidden"
fields << FormContent::Hidden.new(html_node)
when "textarea"
fields << FormContent::Textarea.new(html_node)
else
fields << FormContent::Field.new(html_node)
end
end
end
# Find all textarea tags
@node.css("textarea").each do |node|
node = node.as(Lexbor::Node)
next if node["name"].empty?
@fields << FormContent::Textarea.new(node, node.inner_text)
end
@node.css("button").each do |node|
node = node.as(Lexbor::Node)
type = node.fetch("type", "submit").downcase
next if type == "reset"
@buttons << FormContent::Button.new(node, @node)
end
# Find all select tags
@node.css("select").each do |node|
node = node.as(Lexbor::Node)
next if node["name"].empty?
if node.has_key?("multiple")
@selectboxes << FormContent::MultiSelectList.new(node)
else
@selectboxes << FormContent::SelectList.new(node)
# Find all textarea tags
@node.css("textarea").each do |node|
node = node.as(Lexbor::Node)
next if node["name"].empty?
@fields << FormContent::Textarea.new(node, node.inner_text)
end
end
end
private def build_query_string(params : Array(Array(String)))
params.reduce("") do |acc, arr|
hash = {arr[0] => arr[1]}
acc + URI::Params.encode(hash) + '&'
end.rchop
end
private def build_query
query = [] of Array(String)
successful_controls = Array(FormContent::Field | FormContent::CheckBox).new
fields.each do |elm|
successful_controls << elm
end
checkboxes.each do |elm|
if elm.checked
successful_controls << elm
@node.css("button").each do |node|
node = node.as(Lexbor::Node)
type = node.fetch("type", "submit").downcase
next if type == "reset"
@buttons << FormContent::Button.new(node, @node)
end
end
radio_groups = Hash(String, Array(FormContent::RadioButton)).new
radiobuttons.each do |radio|
name = radio.name
radio_groups[name] = Array(FormContent::RadioButton).new unless radio_groups.has_key?(name)
radio_groups[name] << radio
end
radio_groups.each_value do |g|
checked = g.select(&.checked)
if checked.uniq.size > 1
# values = checked.map(&.value).join(', ').inspect
# name = checked.first.name.inspect
# raise Mechanize::Error,
# "radiobuttons #{values} are checked in the #{name} group, " \
# "only one is allowed"
raise Mechanize::Error.new
else
successful_controls << checked.first unless checked.empty?
end
end
@clicked_buttons.each do |b|
successful_controls << b
end
successful_controls.each do |ctrl|
value = ctrl.query_value
next if value[0] == ""
query.push(value)
end
@selectboxes.each do |s|
value = s.query_value
if value
value.each do |v|
query.push(v)
# Find all select tags
@node.css("select").each do |node|
node = node.as(Lexbor::Node)
next if node["name"].empty?
if node.has_key?("multiple")
@selectboxes << FormContent::MultiSelectList.new(node)
else
@selectboxes << FormContent::SelectList.new(node)
end
end
end
query
end
# This method adds a button to the query. If the form needs to be
# submitted with multiple buttons, pass each button to this method.
def add_button_to_query(button)
unless button.form_node == @node
message = ""
"#{button.inspect} does not belong to the same page as " \
"the form #{@name.inspect} in #{@page.try &.uri}"
message = "not a valid button"
raise ArgumentError.new(message)
private def build_query_string(params : Array(Array(String)))
params.reduce("") do |acc, arr|
hash = {arr[0] => arr[1]}
acc + URI::Params.encode(hash) + '&'
end.rchop
end
@clicked_buttons << button
private def build_query
query = [] of Array(String)
successful_controls = Array(FormContent::Field | FormContent::CheckBox).new
fields.each do |elm|
successful_controls << elm
end
checkboxes.each do |elm|
if elm.checked
successful_controls << elm
end
end
radio_groups = Hash(String, Array(FormContent::RadioButton)).new
radiobuttons.each do |radio|
name = radio.name
radio_groups[name] = Array(FormContent::RadioButton).new unless radio_groups.has_key?(name)
radio_groups[name] << radio
end
radio_groups.each_value do |g|
checked = g.select(&.checked)
if checked.uniq.size > 1
# values = checked.map(&.value).join(', ').inspect
# name = checked.first.name.inspect
# raise Mechanize::Error,
# "radiobuttons #{values} are checked in the #{name} group, " \
# "only one is allowed"
raise Error.new
else
successful_controls << checked.first unless checked.empty?
end
end
@clicked_buttons.each do |b|
successful_controls << b
end
successful_controls.each do |ctrl|
value = ctrl.query_value
next if value[0] == ""
query.push(value)
end
@selectboxes.each do |s|
value = s.query_value
if value
value.each do |v|
query.push(v)
end
end
end
query
end
# This method adds a button to the query. If the form needs to be
# submitted with multiple buttons, pass each button to this method.
def add_button_to_query(button)
unless button.form_node == @node
message = ""
"#{button.inspect} does not belong to the same page as " \
"the form #{@name.inspect} in #{@page.try &.uri}"
message = "not a valid button"
raise ArgumentError.new(message)
end
@clicked_buttons << button
end
end
end

2
src/mechanize/form/button.cr
View File

@ -1,5 +1,5 @@
# This class represents button related html element.
# <button>, and <input> whose type is button, reset, image, submit.
# &lt;button&gt;, and &lt;input&gt; whose type is button, reset, image, submit.
class Mechanize::FormContent::Button < Mechanize::FormContent::Field
getter form_node : Node | Lexbor::Node

2
src/mechanize/form/check_box.cr
View File

@ -1,4 +1,4 @@
# This class represents <input type="checkbox">
# This class represents &lt;input type="checkbox"&gt;
class Mechanize::FormContent::CheckBox < Mechanize::FormContent::RadioButton
# set checkbox checked
def check

2
src/mechanize/form/field.cr
View File

@ -1,4 +1,4 @@
# This class represents <input> elements in the form.
# This class represents &lt;input&gt; elements in the form.
class Mechanize::FormContent::Field
# returns field's 'value' attribute
property value : String?

2
src/mechanize/form/hidden.cr
View File

@ -1,3 +1,3 @@
# This class represents <input type="hidden">
# This class represents &lt;input type="hidden"&gt;
class Mechanize::FormContent::Hidden < Mechanize::FormContent::Field
end

2
src/mechanize/form/image_button.cr
View File

@ -1,3 +1,3 @@
# This class represents <input type="image">
# This class represents &lt;input type="image"&gt;
class Mechanize::FormContent::ImageButton < Mechanize::FormContent::Button
end

2
src/mechanize/form/multi_select_list.cr
View File

@ -1,6 +1,6 @@
require "./option"
# This class represents <select multiple>
# This class represents &lt;select multiple&gt;
class Mechanize::FormContent::MultiSelectList
getter node : Lexbor::Node
getter name : String

1
src/mechanize/form/option.cr
View File

@ -1,3 +1,4 @@
# This class represents &lt;option&gt; of &lt;select&gt;
class Mechanize::FormContent::Option
getter select_list : FormContent::MultiSelectList
getter node : Lexbor::Node

2
src/mechanize/form/radio_button.cr
View File

@ -1,4 +1,4 @@
# This class represents <input type="radio">
# This class represents &lt;input type="radio"&gt;
class Mechanize::FormContent::RadioButton < Mechanize::FormContent::Field
property :checked, :form

2
src/mechanize/form/reset_button.cr
View File

@ -1,3 +1,3 @@
# This class represents <input type="reset">
# This class represents &lt;input type="reset"&gt;
class Mechanize::FormContent::ResetButton < Mechanize::FormContent::Button
end

2
src/mechanize/form/select_list.cr
View File

@ -1,6 +1,6 @@
require "./multi_select_list"
# This class represents <select> which is not multiple
# This class represents &lt;select&gt; which is not multiple
class Mechanize::FormContent::SelectList < Mechanize::FormContent::MultiSelectList
def initialize(node)
super node

2
src/mechanize/form/submit_button.cr
View File

@ -1,3 +1,3 @@
# This class represents <input type="submit">
# This class represents &lt;input type="submit"&gt;
class Mechanize::FormContent::SubmitButton < Mechanize::FormContent::Button
end

2
src/mechanize/form/text.cr
View File

@ -1,3 +1,3 @@
# This class represents <input type="text">
# This class represents &lt;input type="text"&gt;
class Mechanize::FormContent::Text < Mechanize::FormContent::Field
end

2
src/mechanize/form/textarea.cr
View File

@ -1,3 +1,3 @@
# This class represents <input type="textarea">
# This class represents &lt;input type="textarea"&gt;
class Mechanize::FormContent::Textarea < Mechanize::FormContent::Field
end

8
src/mechanize/history.cr
View File

@ -1,17 +1,17 @@
require "./page"
# This class represents the history of http response you sent.
# If you send a request, mechanize saves the history.
class Mechanize
# This class represents the history of http response you sent.
# If you send a request, mechanize saves the history.
class History
# max page size history can save. default is 100.
# as same as `agent.max_history`.
property max_size : Int32
property array : Array(Mechanize::Page)
property array : Array(Page)
delegate :size, :empty?, :last, to: array
def initialize(max_size = 100, array = Array(Mechanize::Page).new)
def initialize(max_size = 100, array = Array(Page).new)
@max_size = max_size
@array = array
end

11
src/mechanize/http/agent.cr
View File

@ -5,15 +5,16 @@ require "../history"
class Mechanize
module HTTP
# This class handles web interation mechanize made.
class Agent
property request_headers : ::HTTP::Headers
property context : Mechanize?
property history : Mechanize::History
property history : History
property user_agent : String
property request_cookies : ::HTTP::Cookies
def initialize(@context : Mechanize? = nil)
@history = Mechanize::History.new
@history = History.new
@request_headers = ::HTTP::Headers.new
@context = context
@request_cookies = ::HTTP::Cookies.new
@ -72,7 +73,7 @@ class Mechanize
# ```
# agent.current_page
# ```
def current_page : Mechanize::Page
def current_page : Page
@history.last
end
@ -80,7 +81,7 @@ class Mechanize
# ```
# agent.back
# ```
def back : Mechanize::Page
def back : Page
@history.pop
end
@ -115,7 +116,7 @@ class Mechanize
end
# Sets a Referer header.
def set_request_referer(referer : Mechanize::Page?)
def set_request_referer(referer : Page?)
return unless referer
request_headers["Referer"] = referer.uri.to_s

12
src/mechanize/page.cr
View File

@ -2,12 +2,12 @@ require "./file"
require "./utils/element_matcher"
require "./page/link"
# This class represents the result of http response.
# If you send a request, it returns the instance of `Mechanize::Page`.
# You can get status code, title, and page body, and search html node using css selector from page instance.
class Mechanize
# This class represents the result of http response.
# If you send a request, it returns the instance of `Mechanize::Page`.
# You can get status code, title, and page body, and search html node using css selector from page instance.
class Page < Mechanize::File
include Mechanize::ElementMatcher
include ElementMatcher
# look at lexbor document.(https://github.com/kostya/lexbor#readme)
delegate :css, to: parser
@ -42,7 +42,7 @@ class Mechanize
# ```
# page.forms # => Array(Mechanize::Form)
# ```
def forms : Array(Mechanize::Form)
def forms : Array(Form)
forms = css("form").map do |html_form|
form = Form.new(html_form, self)
form.action ||= @uri.to_s
@ -54,7 +54,7 @@ class Mechanize
# ```
# page.links # => Array(Mechanize::PageContent::Link)
# ```
def links : Array(Mechanize::PageContent::Link)
def links : Array(PageContent::Link)
links = %w{a area}.map do |tag|
css(tag).map do |node|
PageContent::Link.new(node, @mech, self)

34
src/mechanize/page/link.cr
View File

@ -1,22 +1,24 @@
# This class represents link element like <a> and <area>.
# The instance of this class is clickable.
class Mechanize::PageContent::Link
getter node : Lexbor::Node
getter page : Mechanize::Page
getter mech : Mechanize
getter href : String
getter text : String
class Mechanize
class PageContent::Link
getter node : Lexbor::Node
getter page : Page
getter mech : Mechanize
getter href : String
getter text : String
def initialize(node, mech, page)
@node = node
@page = page
@mech = mech
@href = node.fetch("href", "")
@text = node.inner_text
end
def initialize(node, mech, page)
@node = node
@page = page
@mech = mech
@href = node.fetch("href", "")
@text = node.inner_text
end
# click on this link
def click
@mech.click self
# click on this link
def click
@mech.click self
end
end
end

2
src/mechanize/utils/element_matcher.cr
View File

@ -65,7 +65,7 @@ class Mechanize
def {{singular.id}}_with(criteria)
f = {{plural.id}}_with(criteria)
# TODO: Write correct error message.
raise Mechanize::ElementNotFoundError.new(:{{singular.id}}, "") if f.empty?
raise ElementNotFoundError.new(:{{singular.id}}, "") if f.empty?
f.first
end
end