scinote-web/app/services/reports/docx/private_methods.rb

329 lines
8.6 KiB
Ruby
Raw Normal View History

# frozen_string_literal: true
2019-09-17 16:50:01 +08:00
module Reports::Docx::PrivateMethods
private
# RTE fields support
def html_to_word_converter(text)
html = Nokogiri::HTML(text)
raw_elements = recursive_children(html.css('body').children, [])
# Combined raw text blocks in paragraphs
2020-07-15 15:44:25 +08:00
elements = combine_docx_elements(raw_elements)
# Draw elements
elements.each do |elem|
if elem[:type] == 'p'
2020-07-15 15:44:25 +08:00
Reports::Docx.render_p_element(@docx, elem, scinote_url: @scinote_url, link_style: @link_style)
elsif elem[:type] == 'table'
tiny_mce_table(elem[:data])
elsif elem[:type] == 'newline'
style = elem[:style] || {}
2020-08-05 22:14:43 +08:00
# print heading if its heading
# Mixing heading with other style setting causes problems for Word
if %w(h1 h2 h3 h4 h5).include?(style[:style])
@docx.public_send(style[:style], elem[:value])
else
@docx.p elem[:value] do
align style[:align]
color style[:color]
bold style[:bold]
italic style[:italic]
end
end
elsif elem[:type] == 'image'
2020-07-15 15:44:25 +08:00
Reports::Docx.render_img_element(@docx, elem)
end
end
end
def combine_docx_elements(raw_elements)
elements = []
temp_p = []
raw_elements.each do |elem|
if %w(image newline table).include? elem[:type]
unless temp_p.empty?
elements.push(type: 'p', children: temp_p)
temp_p = []
2019-10-01 16:26:24 +08:00
end
2020-07-15 15:44:25 +08:00
elements.push(elem)
elsif %w(br text a).include? elem[:type]
temp_p.push(elem)
end
end
2020-07-15 15:44:25 +08:00
elements.push(type: 'p', children: temp_p)
elements
end
# Convert HTML structure to plain text structure
2020-07-15 15:44:25 +08:00
def recursive_children(children, elements, options = {})
children.each do |elem|
if elem.class == Nokogiri::XML::Text
2019-07-02 17:21:14 +08:00
next if elem.text.strip == ' ' # Invisible symbol
style = paragraph_styling(elem.parent)
type = (style[:align] && style[:align] != :justify) || style[:style] ? 'newline' : 'text'
text = smart_annotation_check(elem)
elements.push(
type: type,
value: text.strip.delete(' '), # Invisible symbol
style: style
)
next
end
if elem.name == 'br'
elements.push(type: 'br')
next
end
if elem.name == 'img' && elem.attributes['data-mce-token']
2019-10-01 16:26:24 +08:00
image = TinyMceAsset.find_by(id: Base62.decode(elem.attributes['data-mce-token'].value))
next unless image
image_path = image_path(image.image)
dimension = FastImage.size(image_path)
2020-07-15 15:44:25 +08:00
next unless dimension
style = image_styling(elem, dimension)
elements.push(
type: 'image',
2019-10-01 16:26:24 +08:00
data: image_path.split('&')[0],
blob: image.blob,
style: style
)
next
end
if elem.name == 'a'
2020-07-15 15:44:25 +08:00
elements.push(link_element(elem))
next
end
if elem.name == 'table'
elem = tiny_mce_table(elem, nested_table: true) if options[:nested_tables]
elements.push(
type: 'table',
data: elem
)
next
end
elements = recursive_children(elem.children, elements) if elem.children
end
elements
end
2020-07-15 15:44:25 +08:00
def link_element(elem)
text = elem.text
link = elem.attributes['href'].value if elem.attributes['href']
if elem.attributes['class']&.value == 'record-info-link'
link = nil
text = "##{text}"
end
text = "##{text}" if elem.parent.attributes['class']&.value == 'atwho-inserted'
text = "@#{text}" if elem.attributes['class']&.value == 'atwho-user-popover'
{
type: 'a',
value: text,
link: link
}
end
def smart_annotation_check(elem)
return "[#{elem.text}]" if elem.parent.attributes['class']&.value == 'sa-type'
elem.text
end
# Prepare style for text
def paragraph_styling(elem)
style = elem.attributes['style']
result = {}
result[:style] = elem.name if elem.name.include? 'h'
result[:bold] = true if elem.name == 'strong'
result[:italic] = true if elem.name == 'em'
2019-10-01 16:26:24 +08:00
style_keys = %w(text-align color)
if style
style_keys.each do |key|
style_el = style.value.split(';').select { |i| (i.include? key) }[0]
next unless style_el
value = style_el.split(':')[1].strip if style_el
if key == 'text-align'
result[:align] = value.to_sym
elsif key == 'color' && calculate_color_hsp(value) < 190
result[:color] = value.delete('#')
end
end
end
result
end
# Prepare style for images
def image_styling(elem, dimension)
dimension[0] = elem.attributes['width'].value.to_i if elem.attributes['width']
dimension[1] = elem.attributes['height'].value.to_i if elem.attributes['height']
if elem.attributes['style']
align = if elem.attributes['style'].value.include? 'margin-right'
:center
elsif elem.attributes['style'].value.include? 'float: right'
:right
else
:left
end
end
margins = Constants::REPORT_DOCX_MARGIN_LEFT + Constants::REPORT_DOCX_MARGIN_RIGHT
max_width = (Constants::REPORT_DOCX_WIDTH - margins) / 20
if dimension[0] > max_width
x = max_width
y = dimension[1] * max_width / dimension[0]
else
x = dimension[0]
y = dimension[1]
end
{
width: x,
height: y,
2020-07-15 15:44:25 +08:00
align: align,
max_width: max_width
}
end
def asset_image_preparing(asset)
return unless asset
image_path = image_path(asset.file)
dimension = FastImage.size(image_path)
x = dimension[0]
y = dimension[1]
if x > 300
y = y * 300 / x
x = 300
end
2019-10-01 16:26:24 +08:00
@docx.img image_path.split('&')[0] do
data asset.blob.download
width x
height y
end
end
def initial_document_load
@docx.page_size do
width Constants::REPORT_DOCX_WIDTH
height Constants::REPORT_DOCX_HEIGHT
end
@docx.page_margins do
left Constants::REPORT_DOCX_MARGIN_LEFT
right Constants::REPORT_DOCX_MARGIN_RIGHT
top Constants::REPORT_DOCX_MARGIN_TOP
bottom Constants::REPORT_DOCX_MARGIN_BOTTOM
end
@docx.page_numbers true, align: :right
path = Rails.root.join('app', 'assets', 'images', 'logo.png')
@docx.img path.to_s do
height 20
width 100
align :left
end
@docx.p do
text I18n.t('projects.reports.new.generate_PDF.generated_on', timestamp: I18n.l(Time.zone.now, format: :full))
br
end
generate_html_styles
end
def generate_html_styles
@docx.style do
2020-07-08 17:35:25 +08:00
id 'Heading1'
name 'heading 1'
font 'Arial'
size 36
2020-07-08 17:35:25 +08:00
bottom 120
bold true
end
@link_style = {
color: '37a0d9',
bold: true
}
@color = {
gray: 'a0a0a0',
green: '2dbe61'
}
end
2020-07-15 15:44:25 +08:00
def tiny_mce_table(table_data, options = {})
docx_table = []
scinote_url = @scinote_url
link_style = @link_style
table_data.css('tbody').first.children.each do |row|
docx_row = []
next unless row.name == 'tr'
row.children.each do |cell|
next unless cell.name == 'td'
# Parse cell content
formated_cell = recursive_children(cell.children, [], nested_tables: true)
# Combine text elements to single paragraph
formated_cell = combine_docx_elements(formated_cell)
docx_cell = Caracal::Core::Models::TableCellModel.new do |c|
formated_cell.each do |cell_content|
if cell_content[:type] == 'p'
Reports::Docx.render_p_element(c, cell_content,
2020-07-15 15:47:29 +08:00
scinote_url: scinote_url, link_style: link_style, skip_br: true)
2020-07-15 15:44:25 +08:00
elsif cell_content[:type] == 'table'
2020-08-10 19:47:32 +08:00
c.table formated_cell_content[:data], border_size: Constants::REPORT_DOCX_TABLE_BORDER_SIZE
2020-07-15 15:44:25 +08:00
elsif cell_content[:type] == 'image'
Reports::Docx.render_img_element(c, cell_content, table: { columns: row.children.length / 3 })
end
end
end
docx_row.push(docx_cell)
end
docx_table.push(docx_row)
end
if options[:nested_table]
docx_table
else
2020-08-10 19:47:32 +08:00
@docx.table docx_table, border_size: Constants::REPORT_DOCX_TABLE_BORDER_SIZE
2020-07-15 15:44:25 +08:00
end
end
def image_path(attachment)
attachment.service_url
end
def calculate_color_hsp(color)
return 255 if color.length != 7
color = color.delete('#').scan(/.{1,2}/)
rgb = color.map(&:hex)
Math.sqrt(
0.299 * (rgb[0]**2) +
0.587 * (rgb[1]**2) +
0.114 * (rgb[2]**2)
)
end
end