microformats 0.18.2

use crate::parse::{element::Extraction, remove_surrounding_whitespace};

use super::*;
pub struct PropertiesParser {
    item: ElementPtr,
    properties: Properties,
    base_url: Url,
}

trait ImpliedPropertyExtraction {
    fn can_extract(properties: &Properties) -> bool;
    fn extract(self, base_url: &url::Url) -> Option<PropertyValue>;
}

struct ImpliedUrlExtractor(ElementPtr);

const IMPLIED_URL_TAGS: [&str; 2] = ["a", "area"];

impl ImpliedUrlExtractor {
    fn get_first_level_element(&self) -> Option<Node> {
        // if a.h-x[href] or area.h-x[href] then use that [href] for url
        let node = self.0.node.to_owned();
        if node.root_classes().is_empty() {
            return None;
        }

        if !IMPLIED_URL_TAGS.contains(&node.tag()) {
            return None;
        }

        Some(node)
    }

    fn get_second_level_element(&self) -> Option<Node> {
        // [else] if .h-x>a[href]:only-of-type:not[.h-*], then use that [href] for url
        // else if .h-x>area[href]:only-of-type:not[.h-*], then use that [href] for url
        let parent_node = self.0.node.to_owned();

        if parent_node.root_classes().is_empty() {
            return None;
        }

        let child_elems = parent_node.elements();

        let elems: Vec<_> = child_elems
            .iter()
            .filter(|node| IMPLIED_URL_TAGS.contains(&node.tag()) && node.root_classes().is_empty())
            .cloned()
            .collect();

        if elems.len() != 1 {
            None
        } else {
            elems.first().cloned()
        }
    }

    fn get_third_level_element(&self) -> Option<Node> {
        // [else] if .h-x>:only-child:not[.h-*]>a[href]:only-of-type:not[.h-*], then use that [href] for url
        // else if .h-x>:only-child:not[.h-*]>area[href]:only-of-type:not[.h-*], then use that [href] for url
        let node = self.0.node.to_owned();

        if node.root_classes().is_empty() {
            return None;
        }

        let child_elems = node.elements();

        if child_elems.len() != 1 {
            return None;
        }

        let child_elem = &child_elems[0];

        if !child_elem.root_classes().is_empty() {
            return None;
        }

        let elems: Vec<_> = child_elem
            .elements()
            .into_iter()
            .filter(|node| IMPLIED_URL_TAGS.contains(&node.tag()) && node.root_classes().is_empty())
            .collect();

        if elems.len() != 1 {
            return None;
        }

        elems.first().cloned()
    }
}

impl ImpliedPropertyExtraction for ImpliedUrlExtractor {
    #[tracing::instrument(level = "trace", skip(self), ret)]
    fn extract(self, base_url: &url::Url) -> Option<PropertyValue> {
        [
            self.get_first_level_element(),
            self.get_second_level_element(),
            self.get_third_level_element(),
        ]
        .into_iter()
        .flatten()
        .find_map(|node| {
            node.attr("href").map(|href_attr_value| {
                if let Ok(u) = base_url.join(&href_attr_value) {
                    PropertyValue::Url(UrlValue::new(u))
                } else {
                    PropertyValue::Plain(TextValue::new(href_attr_value.to_string()))
                }
            })
        })
        .filter(non_empty_property_value)
    }

    #[tracing::instrument(level = "trace", ret)]
    fn can_extract(properties: &Properties) -> bool {
        !properties.contains_key("url")
            && !properties.values().flatten().any(|property_value| {
                matches!(
                    property_value,
                    PropertyValue::Url(_) | PropertyValue::Image(_)
                )
            })
    }
}

struct ImpliedPhotoExtractor(ElementPtr);

const IMPLIED_PHOTO_TAGS: [&str; 2] = ["img", "object"];

impl ImpliedPhotoExtractor {
    fn get_first_level_element(&self) -> Option<Node> {
        let node = self.0.node.to_owned();
        if node.root_classes().is_empty() {
            return None;
        }

        if !IMPLIED_PHOTO_TAGS.contains(&node.tag()) {
            return None;
        }

        Some(node)
    }

    fn get_second_level_element(&self) -> Option<Node> {
        let parent_node = self.0.node.to_owned();

        if parent_node.root_classes().is_empty() {
            return None;
        }

        let child_elems = parent_node.elements();

        let elems: Vec<_> = child_elems
            .iter()
            .filter(|node| IMPLIED_PHOTO_TAGS.contains(&node.tag()))
            .cloned()
            .collect();

        if elems.len() != 1 {
            None
        } else {
            elems.first().cloned()
        }
    }

    fn get_third_level_element(&self) -> Option<Node> {
        let node = self.0.node.to_owned();

        if node.root_classes().is_empty() {
            return None;
        }

        let child_elems = node.elements();

        if child_elems.len() != 1 {
            return None;
        }

        let child_elem = &child_elems[0];

        let elems: Vec<_> = child_elem
            .elements()
            .into_iter()
            .filter(|node| {
                IMPLIED_PHOTO_TAGS.contains(&node.tag()) && node.root_classes().is_empty()
            })
            .collect();

        if elems.len() != 1 {
            return None;
        }

        elems.first().cloned()
    }
}

impl ImpliedPropertyExtraction for ImpliedPhotoExtractor {
    fn extract(self, base_url: &url::Url) -> Option<PropertyValue> {
        [
            self.get_first_level_element(),
            self.get_second_level_element(),
            self.get_third_level_element(),
        ]
        .into_iter()
        .flatten()
        .find_map(|node| {
            let src_attr_value = extract_img_element(&node, base_url);
            let data_attr_str = node
                .attr("data")
                .filter(non_empty_string)
                .filter(|_| node.tag() == "object")
                .map(|data_attr_value| {
                    if let Ok(u) = base_url.join(&data_attr_value) {
                        PropertyValue::Url(UrlValue::new(u))
                    } else {
                        PropertyValue::Plain(TextValue::new(data_attr_value.to_string()))
                    }
                });

            src_attr_value.or(data_attr_str)
        })
        .filter(non_empty_property_value)
    }

    fn can_extract(properties: &Properties) -> bool {
        !properties.contains_key("photo")
            && !properties.values().flatten().any(|property_value| {
                matches!(
                    property_value,
                    PropertyValue::Url(_)
                        | PropertyValue::Fragment { .. }
                        | PropertyValue::Item(_)
                        | PropertyValue::Image(_)
                )
            })
    }
}

struct ImpliedNameExtractor(ElementPtr);

const IMPLIED_NAME_TAGS: [&str; 3] = ["img", "area", "abbr"];
const IMPLIED_NAME_TAGS_TO_ATTR: [(&str, &str); 3] =
    [("img", "alt"), ("area", "alt"), ("abbr", "title")];

impl ImpliedNameExtractor {
    fn get_first_level_element(&self) -> Option<Node> {
        let node = self.0.node.to_owned();
        if node.root_classes().is_empty() {
            return None;
        }

        if !IMPLIED_NAME_TAGS.contains(&node.tag()) && node.tag() != "area" {
            return None;
        }

        Some(node)
    }

    fn attr_name(desired_tag: &str) -> Option<String> {
        IMPLIED_NAME_TAGS_TO_ATTR
            .iter()
            .find_map(|(tag, attr)| Some(attr.to_string()).filter(|_| tag == &desired_tag))
    }

    fn require_attr_in_node(node: &Node) -> bool {
        IMPLIED_NAME_TAGS.contains(&node.tag()) && node.root_classes().is_empty() && {
            if let Some(attr) = Self::attr_name(node.tag()) {
                node.attr(&attr).filter(non_empty_string).is_some()
            } else {
                false
            }
        }
    }

    fn get_second_level_element(&self) -> Option<Node> {
        let node = self.0.node.to_owned();

        if node.root_classes().is_empty() {
            return None;
        }
        let node_elems = node.elements();

        let elems: Vec<_> = node_elems
            .iter()
            .filter(|&x| Self::require_attr_in_node(x))
            .cloned()
            .collect();

        if node_elems.len() != 1 {
            return None;
        }

        elems
            .first()
            .cloned()
            .filter(|elem| elem.root_classes().is_empty())
    }

    fn get_third_level_element(&self) -> Option<Node> {
        let node = self.0.node.to_owned();

        if node.root_classes().is_empty() {
            return None;
        }

        let elems: Vec<_> = node
            .elements()
            .into_iter()
            .filter(|node| node.root_classes().is_empty())
            .collect();

        if elems.len() != 1 {
            return None;
        }

        let child_elem = &elems[0];
        let child_elems: Vec<_> = child_elem
            .elements()
            .into_iter()
            .filter(Self::require_attr_in_node)
            .collect();

        if child_elems.len() != 1 {
            return None;
        }

        child_elems.first().cloned()
    }
}

impl ImpliedPropertyExtraction for ImpliedNameExtractor {
    fn extract(self, base_url: &url::Url) -> Option<PropertyValue> {
        [
            self.get_first_level_element(),
            self.get_second_level_element(),
            self.get_third_level_element(),
        ]
        .into_iter()
        .flatten()
        .find_map(|node| {
            let attr_name = Self::attr_name(node.tag())?;
            node.attr(&attr_name)
                .map(|s| PropertyValue::Plain(TextValue::new(s)))
        })
        .filter(non_empty_property_value)
        .or_else(|| {
            self.0
                .node
                .text_content(base_url)
                .ok()
                .map(|Extraction { text, .. }| text)
                .map(remove_surrounding_whitespace)
                .map(|s| PropertyValue::Plain(TextValue::new(s)))
                .filter(non_empty_property_value)
        })
    }
    fn can_extract(properties: &Properties) -> bool {
        // if no explicit "name" property, and no other p-* or e-* properties, and no nested microformats,
        !properties.contains_key("name")
            && !properties.values().flatten().any(|stored_value| {
                matches!(
                    stored_value,
                    PropertyValue::Plain(_)
                        | PropertyValue::Fragment { .. }
                        | PropertyValue::Item(_)
                )
            })
    }
}

// TODO: Split this up into a per-implied property parser.
#[allow(dead_code)]
impl PropertiesParser {
    pub fn new(item: ElementPtr, properties: Properties, base_url: &Url) -> Self {
        Self {
            item,
            properties,
            base_url: base_url.to_owned(),
        }
    }

    #[doc(hidden)]
    #[tracing::instrument(level = "trace", skip(self), ret)]
    fn can_imply_url(&self) -> bool {
        !self.properties.contains_key("url")
            && !self.properties.values().flatten().any(|v| {
                matches!(
                    v,
                    PropertyValue::Url(_) | PropertyValue::Fragment { .. } | PropertyValue::Item(_)
                )
            })
    }

    // $ELEM:only-child:not[h-*]
    #[doc(hidden)]
    #[tracing::instrument(level = "trace", skip(elem), ret)]
    fn find_direct_only_child(elem: &swc_html_ast::Element) -> Option<Node> {
        let elems = Node {
            elem: elem.to_owned(),
        }
        .elements();

        if elems.len() != 1 {
            return None;
        }

        let node = elems[0].to_owned();

        if !node.root_classes().is_empty() {
            return None;
        }

        Some(node)
    }

    // *:only-child:not[.h-*]>$ELEM:only-child:not[.h-*]
    #[doc(hidden)]
    #[tracing::instrument(level = "trace", ret)]
    fn find_nested_direct_only_of_type(
        element: &swc_html_ast::Element,
        tag_names: &[&str],
    ) -> Option<Node> {
        let direct_child = Self::find_direct_only_child(element)?;

        tag_names.iter().find_map(|tag_name| {
            direct_child
                .elements()
                .into_iter()
                .filter_map(|node| {
                    if !&node.root_classes().is_empty() {
                        None
                    } else {
                        Some(node)
                    }
                })
                .find(|node| node.tag() == *tag_name)
        })
    }

    #[doc(hidden)]
    #[tracing::instrument(level = "trace", skip(node), ret)]
    fn scan_for_first_of_element(node: &Node, tag_names: &[&str]) -> Option<Node> {
        if tag_names.contains(&node.tag()) {
            Some(node.to_owned())
        } else {
            node.elem
                .children
                .iter()
                .filter_map(|child| {
                    if let Child::Element(elem) = child {
                        Some(Node {
                            elem: elem.to_owned(),
                        })
                    } else {
                        None
                    }
                })
                .find_map(|child_node| {
                    if child_node.mf_classes().is_empty()
                        && child_node.property_classes().is_empty()
                    {
                        Self::scan_for_first_of_element(&child_node, tag_names)
                    } else {
                        None
                    }
                })
        }
    }

    #[doc(hidden)]
    #[tracing::instrument(level = "trace", skip(node), ret)]
    fn imply_url(node: &Node, base_url: &Url) -> Option<PropertyValue> {
        Self::scan_for_first_of_element(node, &["a", "area"])
            .and_then(|node| node.attr("href"))
            .map(|url_like_str| {
                if let Ok(url) = base_url.join(&url_like_str) {
                    PropertyValue::Url(UrlValue::new(url))
                } else {
                    PropertyValue::Plain(TextValue::new(url_like_str))
                }
            })
    }

    #[doc(hidden)]
    #[tracing::instrument(level = "trace", skip(self), ret)]
    fn implied_url(&self) -> Option<PropertyValue> {
        if self.can_imply_url() {
            if let Some(value) = Self::imply_url(&self.item.node, &self.base_url) {
                Some(value)
            } else if let Some(value) = Self::find_direct_only_child(&self.item.node.elem)
                .as_ref()
                .and_then(|node| Self::imply_url(node, &self.base_url))
            {
                Some(value)
            } else {
                Self::find_nested_direct_only_of_type(&self.item.node.elem, &["a", "area"])
                    .as_ref()
                    .and_then(|node| Self::imply_url(node, &self.base_url))
            }
        } else {
            None
        }
    }

    #[tracing::instrument(level = "trace", skip(self), err)]
    pub(crate) fn extract_implied(self) -> Result<Properties, Error> {
        let mut properties = Properties::default();

        if ImpliedNameExtractor::can_extract(&self.properties) {
            if let Some(value) =
                ImpliedNameExtractor(Arc::clone(&self.item)).extract(&self.base_url)
            {
                properties.insert("name".to_string(), vec![value]);
            }
        }

        if ImpliedPhotoExtractor::can_extract(&self.properties) {
            if let Some(value) =
                ImpliedPhotoExtractor(Arc::clone(&self.item)).extract(&self.base_url)
            {
                properties.insert("photo".to_string(), vec![value]);
            }
        }

        if ImpliedUrlExtractor::can_extract(&self.properties) {
            if let Some(value) = ImpliedUrlExtractor(Arc::clone(&self.item)).extract(&self.base_url)
            {
                properties.insert("url".to_string(), vec![value]);
            }
        }

        Ok(properties)
    }
}

#[cfg(test)]
mod test;