def limit_widths(root):
for node in root.iterdescendants():
css_width = utils.get_node_width(node, "px")
width, height = node_size(node)
print node.tag, width, css_width
if css_width > width:
utils.change_node_width(node, "{:.2f}pt".format(width))
def fix_image_tables(root):
img_tables = root.xpath(
'//table[contains(@class, "short-table") and not(contains(@class, "infobox")) and .//a[contains(@class, "image")]]'
)
for table in img_tables:
utils.remove_node_styles(table, "margin")
utils.append_class(table, "image-table")
max_widths = {}
for row in table.xpath(".//tr"):
for n, column in enumerate(row.xpath(".//td")):
for img in column.xpath(".//img"):
width = utils.get_node_width(img, target_unit="px")
max_widths[n] = max(width, max_widths.get(n, 0))
total_width = sum(max_widths.values())
if total_width * config.px2pt > config.page_width_pt:
utils.append_class(table, "wide-image-table")
for row in table.xpath(".//tr"):
for n, column in enumerate(row.xpath(".//td")):
_remove_inner_image_node_width(column, "image")
utils.remove_node_styles(column, ["padding-left", "padding", "margin"])
utils.add_node_style(
column, "width", "{}%".format(max_widths.get(n, 0) / total_width * 100)
)
elif total_width > 0:
for img in table.xpath(".//img"):
_resize_image_node_width_to_pt(img)
def remove_img_style_size(root):
"""
add class to img container and remove explicit width attributes
"""
xpath_conditions = [
'contains(@class,"thumb") ',
'and not(contains(@class, "tmulti"))',
'and not(contains(@class, "thumbinner"))',
'and not(contains(@class, "thumbcaption"))',
'and not(contains(@class, "thumbimage"))',
]
result = root.xpath("//div[{}]".format(" ".join(xpath_conditions)))
for img_container in result:
if "map" in img_container.attrib.get("class", ""):
continue
thumbinner = img_container.xpath('.//*[contains(@class,"thumbinner")]')
for node in thumbinner:
utils.remove_node_styles(node, ["width", "height", "max-width"])
if not img_container.xpath(".//img"):
log.debug("No <img> found in {}".format(etree.tostring(img_container)))
continue
img = img_container.xpath(".//img")[0]
width = utils.get_node_width(img, target_unit="pt")
utils.remove_node_styles(img, ["width", "height"])
cols = int(round(width / (column_width_pt * 4)))
if cols > 3:
cols = 3
cols = cols * 4
utils.append_class(img_container, "col-{}".format(cols))
utils.remove_node_width(img_container)
utils.remove_node_width(img)
def fix_abspos_overlays(root):
for container in root.xpath(
('//*[contains(@style, "position")' ' and contains(@style, "relative")]')
):
w = utils.get_node_width(container, target_unit="px")
h = utils.get_node_height(container, target_unit="px")
if not (w and h):
img = container.xpath(".//img")
if not img:
continue
img = img[0]
w, h = get_img_size(img)
for node in container.xpath(
('.//*[contains(@style, "position")' ' and contains(@style, "absolute")]')
):
style = utils.get_node_style(node)
left = style.get("left")
top = style.get("top")
for attr in ["left", "top"]:
val = locals()[attr]
if not val:
continue
if val.endswith("%"):
continue
elif val.endswith("px"):
val = val[:-2]
elif val.isdigit():
pass
else:
continue
try:
new_val = 100 * int(float(val)) / (w if attr == "left" else h)
except (ValueError, ZeroDivisionError):
continue
utils.add_node_style(node, attr, "{}%".format(new_val))
def limit_size(root):
for img in root.xpath('//img[not(contains(@class, "inline"))]'):
w, h = get_img_size(img)
if w == 0 or h == 0:
continue
if isinstance(h, unicode) or isinstance(h, str):
continue
in_table = any(node.tag == "table" for node in img.iterancestors())
max_height_outside = 7.5 * config.cm2px
max_height_in_table = 5 * config.cm2px
max_height = max_height_in_table if in_table else max_height_outside
max_width = 6.03 * config.cm2px
# downscale if dimensions too big
scale_factor = min(min(1, max_height / h), min(1, max_width / w))
# upscale image too full width, if the image is wider than 70% of max width
if scale_factor == 1:
scaled_height = max_width / w * h
if 0.7 * max_width < w < max_width and scaled_height < max_height:
scale_factor = max_width / w
if scale_factor != 1:
img.set("width", str(w * scale_factor))
img.set("height", str(h * scale_factor))
for node in [n for n in img.iterancestors()]:
w = utils.get_node_width(node, "px")
h = utils.get_node_height(node, "px")
if w:
utils.change_node_width(node, w * scale_factor, unit="px")
if h:
utils.change_node_height(node, h * scale_factor, unit="px")
def get_img_size(node):
"""
get size of a node in px
"""
try:
width = utils.get_node_width(node)
height = utils.get_node_height(node)
return width, height
except TypeError:
return 0, 0
def _resize_image_node_width_to_pt(node):
"""
resize images from px to pt: 96px -> 72pt = shrink to 75%
the scale factor is more or less deliberate but looks decent in sample pages
"""
if node.tag != "img":
return
width = utils.get_node_width(node, target_unit="px")
utils.remove_node_styles(node, ["width", "height"])
utils.remove_node_width(node)
utils.add_node_style(node, "width", "{}px".format(width * config.px2pt))
def _remove_inner_image_node_width(node, inner_class="thumbinner"):
"""
remove explicit widths from an image node
Side effect: removes the node if it doesn't contain an image!
:param node:
:param inner_class: "thumbinner" or "thumbimage"
:return: original width of the image in pt
"""
utils.remove_node_styles(node, ["width", "height", "max-width"])
wrapper_nodes = node.xpath('.//*[contains(@class,"{}")]'.format(inner_class))
for wrapper_node in wrapper_nodes:
utils.remove_node_styles(wrapper_node, ["width", "height", "max-width"])
if not node.xpath(".//img"):
log.debug("No <img> found in {}. Removing node.".format(etree.tostring(node)))
utils.remove_node(node)
return 0
img = node.xpath(".//img")[0]
width = utils.get_node_width(img, target_unit="pt")
utils.remove_node_styles(img, ["width", "height"])
utils.remove_node_width(img)
return width