library(ggplot2)
library(dplyr)
library(rvest)
library(scales)

rows <- read_xml('mom_texts.xml') %>% xml_nodes('sms')
people <- c("Mom", "Kyle")
df <- data.frame(
  ID = rows %>% xml_attr("type"),
  date = rows %>% xml_attr("readable_date"),
  body = rows %>% xml_attr("body")
)

# mutate(df, datetime = strptime(date, "%b %d, %Y %H:%M:%S %p"))

heat_by_day <- ggplot(data = df, aes(as.Date(date, "%b %d, %Y %H:%M:%S %p"), ID)) +
  geom_bin2d(binwidth = c(1, 1)) +
  scale_x_date(breaks = date_breaks("weeks")) +
  theme_solarized_2() +
  theme(axis.text.x = element_text(angle = 75, hjust = 1)) +
  labs(title = "Text messages per day, per person",
       x = "Date",
       y = "Person")

heat_by_day +
#  geom_rect(mapping = aes(xmin = as.Date('Apr 21, 2018 12:00:00', "%b %d, %Y %H:%M:%S") + 0.5, ymin = 0.5,
#                xmax = as.Date('Apr 22, 2018 16:00:00', "%b %d, %Y %H:%M:%S") + 16/24, ymax = 2.5),
#                color = 'gray42', fill = 'forestgreen', alpha = 0.7) +
  scale_y_discrete(labels = people)


df <- df %>% mutate(char_length = nchar(as.character(body)))

text_lengths_by_datetime <- ggplot(data = df %>% group_by(ID), aes(strptime(date, "%b %d, %Y %H:%M:%S %p"), df$char_length, color = df$ID)) +
  geom_point() +
  geom_smooth() +
  scale_x_datetime(breaks = date_breaks("weeks")) +
  theme_solarized_2() +
  theme(axis.text.x = element_text(angle = 75, hjust = 1)) +
  labs(title = "Text Lengths by Date",
       x = "Date and Time",
       y = "Character Length")

text_lengths_by_datetime +
  scale_color_discrete(
    name = "Person",
    breaks = c("1", "2"),
    labels = people
    )

violin_char_length <- ggplot(data = df %>% group_by(ID), aes(ID, char_length, fill = ID)) +
  geom_violin() +
  geom_dotplot(binaxis = "y", stackdir = "center", binwidth = 3) +
  theme_solarized_2() +
  labs(title = "Distribution of Text Lengths",
       x = "Person",
       y = "Character Length")

violin_char_length +
  scale_x_discrete(labels = people) +
  theme(legend.position = "none")